Abstract: Various aspects of this disclosure relate to determining mapping issues in object relational mapping (ORM). An artificial intelligence (AI) model may be trained to identify errors in mapping between relational databases and objects during code compilation. Multiple AI models may be used, with different models being associated with different programming frameworks, thereby making this technique framework agnostic.

Abstract: Systems, methods, and machine-readable media are disclosed for enabling high-performance programming via a composite programming language that gives programmers complete control over the compilation process. The composite programs include two language levels: an object program level (source code), and a metaprogram level that describes how a compiler should be customized in order to optimize the source code for a target hardware environment. When an augmented compiler receives a composite program, it recognizes the metaprogram and implements the one or more parameters specified within the composite program to optimize the compiler for a given target. Once the augmented compiler has been, it proceeds with compiling the source code included in the composite program. The compiled code is then output as machine language and may be executed by one or more computing systems.

Abstract: A system and method may provide assistance to programmers during programming to reduce the number of routine tasks that must be performed. In some aspects, the system may provide for searching a corpus of source code based on keyword or natural language search input. Search results including code entities and snippets of code that are described by the search input are then provided as search results. Some embodiments relate to using a neural network encoder to generate tensor embeddings of source code and related text in a joint tensor space. Relatedness between embeddings in this joint tensor space for text and associated source code is used in some embodiments to facilitate code search.

Abstract: Methods and systems are disclosed that automate and institutionalize many aspects of the process of creating software. Embodiments automate aspects of pricing, software creation, and delivery using a manufacturing-styled approach to development that reuses existing code and other existing software design features.

Abstract: In one example, a method performed by a compiler comprises: receiving a dataflow graph of a neural network, the neural network comprising a neural network operator; receiving information of computation resources and memory resources of a neural network hardware accelerator intended to execute the neural network operator; determining, based on the dataflow graph, iterations of an operation on elements of a tensor included in the neural network operator; determining, based on the information, a mapping between the elements of the tensor to addresses in the portion of the local memory, and a number of the iterations of the operation to be included in a batch, wherein the number of the iterations in the batch are to be executed in parallel by the neural network hardware accelerator; and generating a schedule of execution of the batches of the iterations of the operations.

Abstract: Certain embodiments of the present disclosure provide techniques for performing performance tests against services in a computing environment. The method generally includes deploying application code to an application namespace hosted on a first set of resources in the computing environment. Testing code is deployed to an infrastructure namespace hosted on a second set of resources in the computing environment. A request to test the application code is received. The request generally includes information identifying a load to be generated in testing the application code. A plurality of container instances implementing the test code are instantiated based on the identified load to be generated to test the application code. A test is executed against the application code through the instantiated plurality of container instances.

Abstract: Methods, computer program products, and systems are presented.

Abstract: Techniques are described herein for translating source code in one programming language to source code in another programming language using machine learning. A method includes: receiving first source code in a first higher-level programming language; processing the first source code, or an intermediate representation thereof, using a sequence-to-sequence neural network model to generate a sequence of outputs, each including a probability distribution; generating second source code in a second higher-level programming language by, for each output in the sequence of outputs: determining a highest probability in the probability distribution associated with the output; in response to the highest probability exceeding a first threshold, generating a predicted portion of the second source code based on a token that corresponds to the highest probability; and in response to the highest probability not exceeding the first threshold, generating a placeholder; and outputting the second source code.

Abstract: Implementations are described herein for identifying related source code edits to perform, or to aid in the performance of, various programming tasks. In various implementations, a first edit made to a first source code snippet in a source code editor may be detected. Based on the first edit, a second source code edit to be made to a second source code snippet may be identified. The identifying may include: traversing one or more graphs to determine one or more edge sequences between nodes corresponding to the first and second source code snippets, comparing the one or more edge sequences to a plurality of reference edge sequences between nodes corresponding to historical co-occurrences of the first and second code edits, and identifying the second edit based on the comparing. The source code editor may provide output that includes a recommendation to implement the second edit.

Abstract: Vectorization and scalarization of methods are provided. A plurality of node webs is constructed based on traversing an intermediate representation of a program. Transitive closure of the plurality of node webs is performed to form a set of final node webs. It is determined that each respective node in the set of final node webs can be converted into one of vector operation code or a sequence of scalar operation codes based on at least one node including a specified vector length and only one vector length value being specified within the set of final node webs. Each respective node in the set of final node webs is converted into one of corresponding vector operation code or a corresponding sequence of scalar operation codes to accelerate execution of supported and unsupported methods of the program.

Abstract: Systems, methods, and non-transitory computer readable media are provided for facilitating improved defect resolution. Defect information and defect criteria information may be obtained. The defect information may identify defects of software and/or hardware in development. The defect criteria information may define one or more criteria for measuring the defects. The defects may be measured based on the one or more criteria. A defect analysis interface may be provided. The defect analysis interface may list a limited number of the defects based on the measurements of the defects. The defect analysis interface may provide costs (e.g., computing resources, time, personnel) of solving the defects.

Abstract: Reinforcement learning enables a framework of information technology assets that include software elements, computational hardware assets, and/or, bundled software and computational hardware systems and products. The performance of successive sessions of an inner loop reinforcement learning is directed and monitored by an outer loop reinforcement learning wherein the outer loop reinforcement learning is designed to reduce financial costs and computational asset requirements and/or optimize learning time in successive instantiations of inner loop reinforcement learning training sessions. The framework enables consideration of the license costs of domain specific simulators, the usage cost of hardware platforms, and the progress of a particular reinforcement learning training. The framework further enables reductions of these costs to orchestrate and train a neural network under budget constraints with respect to the available hardware and software licenses available at runtime.

Abstract: Through the systems and methods described herein for provisioning a software automation and tracking performance of the software automation throughout its lifecycle. A realized benefit of deployment of the software automation can be determined and automatically reported according to a schedule. The reports may be provided to certain specified recipients such as project managers, executive officers, sales and/or vendor relations managers, and the like for analysis and processing by the various parties associated with the operation of the software automation. This all-in-one system provides a platform from which one or more software automation projects may be automatically managed through completion and deployment, improving efficiency of the project and management of all deployed software automations for a more cost-effective suite of such programs.

Abstract: To improve the technological process of programming a computer using a dynamic programming language, generate a first portion of training data which maps types in the dynamic programming language to corresponding functions and methods by performing information retrieval on documentation libraries in the dynamic programming language and/or generate a second portion of training data which maps program variables to the corresponding functions and methods by performing data flow analysis on a plurality of pre-existing programs written in the dynamic programming language. Train a neural network on the first and/or second portions of training data to infer unknown types in the dynamic programming language. Carry out inference with the trained neural network to infer the unknown types. Facilitate programming in the dynamic programming language based on the inferred unknown types. Optionally, execute a resulting program.

Abstract: An order is received indicating a network service model. A context of the order is identified. A deployment plan is generated using the network service model, the deployment plan facilitating an instantiation of a contextually-motivated network service instance as a set of normalized lifecycle management (LCM) operations performed against each of a plurality of associated service entities. The deployment plan is deployed, the deploying including binding each of the normalized LCM operations, based on the context of the order, to one or more respective micro-capabilities, each of the respective micro-capabilities having previously been onboarded to the system as one or more corresponding modeled objects capable of being declaratively composed, each of the corresponding modeled objects including a mapping of object properties, object behaviors, and standard LCM operations to one or more existing micro-capabilities of the system.

Abstract: According to one embodiment, a code information storage is configured to store information on a plurality of functions, a calling relationship between the plurality of functions, and code blocks in the respective plurality of functions. A user operation storage is configured to store information of an already read code block. An already read graph generator is configured to generate an already read graph expressing an already read function including the already read code block, all the code blocks included in the already read function, and a calling relationship between the already read code block and the already read function. An expression of the already read code block differs from an expression of the other code block.

Abstract: A system for developing software provides a graphical user interface on a display of a client device, the graphical user interface displaying features from a library of features for a custom software application, implements simulations of a plurality of the features available for demonstration through the graphical user interface, stores blocks of source code for each feature in a source code repository wherein the blocks are adapted to provide an actual application when compiled by developers, receives from the client device, by a server running a software building component, one or more selected features for the software application, automatically integrates, by the software building component, the one or more selected features to generate an integrated feature set based on attributes of each of the selected features and an inter-feature rules set, and generates an interactive visualization of a navigable prototype of the software application based on the integrated feature set.

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.