Abstract: This document relates to automated generation of machine learning models, such as neural networks. One example method involves obtaining a first machine learning model having one or more first inference operations. The example method also involves identifying a plurality of second inference operations that are supported by an inference hardware architecture. The example method also involves generating second machine learning models by modifying the first machine learning model to include individual second inference operations that are supported by the inference hardware architecture. The example method also involves selecting a final machine learning model from the second machine learning models based on one or more metrics.

Abstract: In examples, a declaration of an ML model is identified within source code of a software project. As a result, a model wrapper may be generated for the ML model and used when compiling and/or executing the software code. Further, a representative object may be generated to enable management of the ML model during the software development process. As an example, model attributes associated with the ML model may be identified from the software code and used to manage the ML model accordingly. In examples, a runtime library associated with the ML model may be automatically included in the software project and/or training of the ML model may be automatically initiated. In some instances, a placeholder ML model or a partially trained or intermediate ML model may be used when building and executing the software project while the ML model is still being trained, thereby enabling continued software development.

Abstract: Generally discussed herein are devices, systems, and methods for improving architecture search and identification with constraints. A method can include receiving, at a compute device, a request for a transformer-based autoregressive language model (TBALM), the request specifying a maximum latency, identifying TBALM architectures that satisfies the maximum latency, identifying a TBALM architecture of the identified TBALM architectures that has a greatest number of decoder parameters resulting in an identified TBALM architecture, and providing the identified TBALM architecture.

Abstract: This document relates to automated generation of machine learning models, such as neural networks. One example system includes a hardware processing unit and a storage resource. The storage resource can store computer-readable instructions cause the hardware processing unit to perform an iterative model-growing process that involves modifying parent models to obtain child models. The iterative model-growing process can also include selecting candidate layers to include in the child models based at least on weights learned in an initialization process of the candidate layers. The system can also output a final model selected from the child models.

Abstract: This document relates to automated generation of machine learning models, such as neural networks. One example system includes a hardware processing unit and a storage resource. The storage resource can store computer-readable instructions cause the hardware processing unit to perform an iterative model-growing process that involves modifying parent models to obtain child models. The iterative model-growing process can also include selecting candidate layers to include in the child models based at least on weights learned in an initialization process of the candidate layers. The system can also output a final model selected from the child models.

Abstract: This document relates to automated generation of machine learning models, such as neural networks. One example system includes a hardware processing unit and a storage resource. The storage resource can store computer-readable instructions cause the hardware processing unit to perform an iterative model-growing process that involves modifying parent models to obtain child models. The iterative model-growing process can also include selecting candidate layers to include in the child models based at least on weights learned in an initialization process of the candidate layers. The system can also output a final model selected from the child models.

Abstract: Examples of the present disclosure relate to generating optimal scanning trajectories for 3D scenes. In an example, a moveable camera may gather information about a scene. During an initial pass, an initial trajectory may be used to gather an initial dataset. In order to generate an optimal trajectory, a reconstruction of the scene may be generated based on the initial data set. Surface points and a camera position graph may be generated based on the reconstruction. A subgradient may be determined, wherein the subgradient provides an additive approximation for the marginal reward associated with each camera position node in the camera position graph. The subgradient may be used to generate an optimal trajectory based on the marginal reward of each camera position node. The optimal trajectory may then be used by to gather additional data, which may be iteratively analyzed and used to further refine and optimize subsequent trajectories.

Abstract: Examples of the present disclosure relate to generating optimal scanning trajectories for 3D scenes. In an example, a moveable camera may gather information about a scene. During an initial pass, an initial trajectory may be used to gather an initial dataset. In order to generate an optimal trajectory, a reconstruction of the scene may be generated based on the initial data set. Surface points and a camera position graph may be generated based on the reconstruction. A subgradient may be determined, wherein the subgradient provides an additive approximation for the marginal reward associated with each camera position node in the camera position graph. The subgradient may be used to generate an optimal trajectory based on the marginal reward of each camera position node. The optimal trajectory may then be used by to gather additional data, which may be iteratively analyzed and used to further refine and optimize subsequent trajectories.

Abstract: Various technologies pertaining to dynamically identifying travel segments to be taken by a traveler traveling in a region are described herein, where observations about travel segments in the region are sparse and subject to alteration. A computer-implemented graph can be loaded into a memory, where the computer-implemented graph is representative of the region. The computer-implemented graph includes nodes that represent locations in the region and edges that represent travel segments of the region, where the edges have costs assigned thereto, and further where there is a defined statistical relationship between the costs. When an observation about a travel path is received, using the computer-implemented graph, inferences can be made about costs of traversing other travel paths in the region.

Abstract: Various technologies pertaining to dynamically identifying travel segments to be taken by a traveler traveling in a region are described herein, where observations about travel segments in the region are sparse and subject to alteration. A computer-implemented graph can be loaded into a memory, where the computer-implemented graph is representative of the region. The computer-implemented graph includes nodes that represent locations in the region and edges that represent travel segments of the region, where the edges have costs assigned thereto, and further where there is a defined statistical relationship between the costs. When an observation about a travel path is received, using the computer-implemented graph, inferences can be made about costs of traversing other travel paths in the region.