Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for interleaving matrix operations of a gated activation unit. One of the methods includes receiving a plurality of weight matrices of a gated activation unit of the neural network, the gated activation unit having two or more layers, each layer defining operations comprising: (i) a matrix operation between a weight matrix for the layer and concatenated input vectors and (ii) a nonlinear activation operation using a result of the matrix operation. Rows of the plurality of weight matrices are interleaved by assigning groups of corresponding rows to respective thread blocks, each thread block being a computation unit for execution by an independent processing unit of a plurality of independent processing units of a parallel processing device.

Abstract: The technology disclosed herein enables computing node allocation based on build process specifications. An example method comprises receiving operational resource requirements of a software build job in a waiting state; identifying, for each computing node of a plurality of computing nodes associated with one or more other software build jobs, a respective set of operational resources installed on the computing node; identifying, among the plurality of computing nodes, a computing node having a minimal, among the plurality of computing nodes, difference between a set of operational resources installed in the computing node and the operational resource requirements of the software job; and scheduling the software build job to execute on the computing node.

Abstract: Methods, storage media, and systems for translating a software expression from a user application programming interface (API) call to an API call of a software development kit (SDK) are disclosed. Some examples may include: receiving a tagged expression indicating that a translation of the software expression from a user API call to an API call of an SDK is to be performed, the SDK being associated with a cloud-native high-performance computing environment, processing an abstract syntax tree associated with the software expression, the processing including replacing symbols in the abstract syntax tree with respective variables, replacing a return statement in the abstract syntax tree with a serialization instruction to write a result to local storage, and serializing the processed abstract syntax tree and providing the serialized abstract syntax tree and one or more resource files to the cloud-native high-performance computing environment for execution.

Abstract: Automatically generating code and related artifacts such as application programming interfaces (APIs) and related documentation from an abstract model of a database. The abstract model is derived from a physical model which may be a source such as a legacy database, an entity relationship diagram, or other schema defining the data tables, objects, entities, or relationships etc. of the source. The generated code may be exposed (that is, made visible to the developer in its pre-compiled state) and further configurable and extendable. Any such extended code is maintained separately from generated code. An API and related documentation are also generated from the same abstract model.

Abstract: Methods and apparatuses are provided for data processing. The method includes receiving a first data packet and a second data packet; associating first codes with the first data packet and second codes with the second data packet to generate a combined data packet after receiving the first data packet and the second data packet, wherein the first codes and the second codes specify processing to be performed to the a combined data packet; generating the combined data packet comprising the first data packet and the second data packet in response to determining that the first data packet and the second data packet are correlated; and performing the processing to the combined data packet in accordance with the first codes or the second codes.

Abstract: In a computer comprising multiple processing units, a method of exchanging read only elements between the processing units is described. The read only elements may be code or data, such as vector or matrix data for an AI graph. A master processing unit is identified. At compile time, at least one shareable read only element is allocated to the master processing unit. The at least one shareable read only element is stored in the local memory of the master processing unit. At compile time a transmitting exchange code sequence designated to be executed at the execution stage of the master processing unit is also allocated to the master processing unit. At a time point determined at compile time, the transmitting exchange code sequence causes the processing unit to identify the shareable read only element and to generate a message to be transmitted for reception by another processing unit, the message comprising the shareable read only data element.

Abstract: Techniques for visualizing dependent relationships in computer program analysis trace elements are described. A code analysis service analyzes computer code associated with a software application for at least one of errors or policy violations. The service outputs a trace log identifying a path within the computer code that results in an error or policy violation. The trace log includes a set of trace elements each corresponding to a line of the computer code. The output is displayed as a tree data structure including nodes. The nodes may include indentations indicating dependent relationships between and among variables initialized at lines in the computer code corresponding to the nodes. Alternatively, the nodes may be arranged as a hierarchy with edges connecting adjacent nodes, and nodes at different levels indicating dependent relationships. The tree data structure may include annotations providing additional contextual information about the actions performed at each node.

Abstract: An architectural diagram recommendation engine is implemented in a data processing system for software architectural diagram analysis and recommendation. The architectural diagram recommendation engine analyzes a software requirements specification document using natural language processing to identify functional requirements and security requirements. The architectural diagram recommendation engine analyzes a digital software architectural diagram image to identify functional components and identifies one or more discrepancies between the functional components of the digital software architectural diagram image and the functional requirements or security requirements. The architectural diagram recommendation engine generates an alert concerning the one or more discrepancies and presents the alert in association with the digital software architectural diagram image.

Abstract: Evolutionary learning techniques are used to validate and prioritize open source software libraries for subsequently determining the best open source software library for a specified technical project. Data associated with the open source software candidates is collected into a cluster and, at an eligibility layer, a fitness score is determined for each of the open source software candidate. Candidates that are determined to meet a required fitness score threshold are passed to the crossover layer, at which, software and hardware standards rules are applied to the open source software metadata to validate the open source software. Invalid candidates are held in queue and subjected to rework analysis. A mutation layer executes the crossover layer iteratively until a predetermined volume of open source candidates results. A ranking layer provides a prioritized ranking list, based on the fitness score, of those open source software candidates that have validated.

Abstract: A chatbot designer user interface (UI) has a visual authoring canvas that shows visual elements indicative of a logical flow of processing in a chatbot application. The designer UI receives an authoring input identifying a language processing system that is to be used during runtime. The chatbot application is configured to use the identified language processing system.

Abstract: Technologies are described for verifying the integrity of data transmitted between a firmware and a baseboard management controller (“BMC”). During a first verification phase, a data integrity tool compares system inventory data provided by firmware protocols to a system inventory module to system inventory data in a device instance format. During a second verification phase, the data integrity tool verifies operation of a firmware conversion module by comparing the system inventory data in the device instance format to the system inventory data in a JavaScript Object Notation (“JSON”) format. During a third verification phase, the data integrity tool retrieves the system inventory data from a management server and compares the system inventory data retrieved from the management server to the system inventory data previously transmitted to the management server. Identified discrepancies can be utilized to modify the system inventory module, the firmware conversion module, or the management server.

Abstract: Proposed are a method and apparatus for accelerating data processing in a neural network. The apparatus for accelerating data processing in a neural network may include: a control unit configured to quantize data by at least one method according to a characteristic of data calculated at a node forming at least one layer constituting the neural network, and to separately perform calculation at the node according to the quantized data; and memory configured to store the quantized data.

Abstract: Techniques herein enable a tenant of a multi-tenant database to select a programming language to interact with a platform that uses a default programming language. A tenant-specific engine may manage a runtime context associated with the tenant in which a tenant may input code that is translated into the default programming language and executed. During execution, the tenant-specific engine may enforce various multi-tenant protections associated with the tenant. For example, the tenant-specific engine may monitor the runtime context and operations of the translated code, and may enforce computational limitations of the tenant as the translated code is executed.

Abstract: A drug research and development software repository and a software package management system. The drug research and development software repository includes a user authentication module, a software package upload and index creation module, a baseline release management module, an authorization management module, a software package retrieval and download module, an adapter plug-in, and a storage layer. The software package management system includes an authentication module, a package management service module, a web front-end module, and a command line tool module.

Abstract: An apparatus includes at least one memory configured to store a virtualization associated with a programming and test platform and a munition. The virtualization includes a virtual machine containing first software or firmware instructions associated with the programming and test platform and second software or firmware instructions that control behavior of the programming and test platform relative to the munition. The virtualization also includes a virtual representation of the munition. The virtualization further includes one or more virtual communication channels configured to communicatively couple the virtual machine and the virtual representation of the munition. The apparatus also includes at least one processor configured to execute the virtualization and simulate the programming and test platform and the munition on the apparatus.

Abstract: A method, system, and computer program product for implementing automated system capacity optimization is provided. The method includes retrieving from plug-in components running on a plurality of hardware and software sources, metrics data associated with the plug-in components. The metrics data is cross-referenced with respect to operational sizing recommendations for each plug-in component based on aggregated disparate sizing guidelines and resulting software code modules are generated. Software and hardware requirements for enabling target computing components are determined based on results of executing the software code modules and operational functionality of the target computing components are enabled in accordance with the software and hardware requirements.

Abstract: Techniques for program verification are described. An exemplary method includes receiving a request to evaluate code based on a customized rule, the customized rule comprising one or more conditions for which the customized rule is applicable and one or more postconditions to indicate at least one check to perform for a given node in a graph for the code, wherein an application of the customized rule performs one or more of: an interleave between a backward analysis and forward analysis based on user-specified conditions, an analysis between sub-graphs by a query from a first sub-graph to a second sub-graph, and an operation on a sub-graph, storage of a result of the operation on the sub-graph, and usage of the stored result in a subsequent operation; generating a graph for the code; and evaluating the code by applying the customized rule to the generated graph.

Abstract: Methods and systems for dynamically updating machine learning models such as text classifiers. One of the methods includes: receiving first data; producing a first machine learning model using the first data; releasing the first machine learning model for use; receiving second data after receipt of the first data; determining that the second data has a difference metric relative to the first data that exceeds a difference threshold; retraining the first machine learning model using at least part of the second data, the retraining producing a second machine learning model; and releasing the second machine learning model for use.

Abstract: A virtual machine that includes a plurality of processes executes on a computer processor. A record-replay file, trace annotations, and an application program interface request are received into the computer processor. The trace annotations and application program interface request are translated into record-replay commands. The record-replay commands capture data from the record-replay file, and the captured data can be accessed via a programmatic interface.

Abstract: A method in a mainframe computing system to invoke a partner product exit routine in an information management system while the information management system is in operation. The method includes scheduling an interrupt routine for execution, creating, by the interrupt routine, an information management system task, where the information management system task is a work unit that provides a logical service within the information management system, scheduling, by the interrupt routine, the information management system task for execution, invoking, by the information management system task, the partner product exit routine, and installing, by the partner product exit routine, a component of a software product that allows the software product to integrate with the information management system.