公开(公告)号：US20230325670A1

公开(公告)日：2023-10-12

申请号：US17820780

申请日：2022-08-18

申请人： NVIDIA Corporation

发明人： Jason Lavar Clemons ,  Stephen W. Keckler ,  Iuri Frosio ,  Jose Manuel Alvarez Lopez ,  Maying Shen

IPC分类号： G06N3/08

CPC分类号： G06N3/082

摘要： A technique for dynamically configuring and executing an augmented neural network in real-time according to performance constraints also maintains the legacy neural network execution path. A neural network model that has been trained for a task is augmented with low-compute “shallow” phases paired with each legacy phase and the legacy phases of the neural network model are held constant (e.g., unchanged) while the shallow phases are trained. During inference, one or more of the shallow phases can be selectively executed in place of the corresponding legacy phase. Compared with the legacy phases, the shallow phases are typically less accurate, but have reduced latency and consume less power. Therefore, processing using one or more of the shallow phases in place of one or more of the legacy phases enables the augmented neural network to dynamically adapt to changes in the execution environment (e.g., processing load or performance requirement).

公开(公告)号：US20170004089A1

公开(公告)日：2017-01-05

申请号：US14788593

申请日：2015-06-30

申请人： NVIDIA Corporation

发明人： Jason Lavar Clemons ,  Chih-Chi Cheng ,  Daniel Robert Johnson ,  Stephen William Keckler ,  Iuri Frosio ,  Yun-Ta Tsai

IPC分类号： G06F12/08

CPC分类号： G06F12/0893 ,  G06F2212/454 ,  G06F2212/455

摘要： A patch memory system for accessing patches from a memory is disclosed. A patch is an abstraction that refers to a contiguous, array of data that is a subset of an N-dimensional array of data. The patch memory system includes a tile cache, and is configured to fetch data associated with a patch by determining one or more tiles associated with an N-dimensional array of data corresponding to the patch, and loading data for the one or more tiles from the memory into the tile cache. The N-dimensional array of data may be a two-dimensional (2D) digital image comprising a plurality of pixels. A patch of the 2D digital image may refer to a 2D subset of the image.

摘要翻译： 公开了一种用于从存储器访问补丁的补丁存储器系统。 补丁是指一个连续的数据数组的抽象，它是N维数据数组的子集。 所述补丁存储器系统包括瓦片高速缓存，并且被配置为通过确定与对应于所述补丁的N维数组阵列相关联的一个或多个瓦片来提取与补丁相关联的数据，以及将所述一个或多个瓦片的数据从 内存进入瓦片缓存。 N维数据阵列可以是包括多个像素的二维（2D）数字图像。 2D数字图像的补丁可以指图像的2D子集。

公开(公告)号：US20240087333A1

公开(公告)日：2024-03-14

申请号：US17943576

申请日：2022-09-13

申请人： NVIDIA Corporation

发明人： Siva Kumar Sastry Hari ,  Jason Lavar Clemons ,  Timothy Kohchih Tsai

IPC分类号： G06V20/58 ,  G06V10/774 ,  G06V10/776 ,  G06V10/82

CPC分类号： G06V20/58 ,  G06V10/774 ,  G06V10/776 ,  G06V10/82 ,  G06V2201/08

摘要： In various examples, techniques for detecting occluded objects within an environment are described. For instance, systems and methods may receive training data representing images and ground truth data indicating whether the images are associated with occluded objects or whether the images are not associated with occluded objects. The systems and methods may then train a neural network to detect occluded objects using the training data and the ground truth data. After training, the systems and methods may use the neural network to detect occluded objects within an environment. For instance, while a vehicle is navigating, the vehicle may process sensor data using the neural network. The neural network may then output data indicating whether an object is located within the environment and occluded from view of the vehicle. In some examples, the neural network may further output additional information associated with the occluded object.

公开(公告)号：US20230111375A1

公开(公告)日：2023-04-13

申请号：US17724819

申请日：2022-04-20

申请人： NVIDIA Corporation

发明人： Jason Lavar Clemons ,  Kavya Sreedhar ,  Stephen W. Keckler

IPC分类号： G06N3/08 ,  G06N3/04 ,  G06F11/34

摘要： A neural network model is augmented for dynamic configuration and execution in real-time according to performance constraints. In an embodiment, the neural network model is a transformer neural network model. The performance constraints may include a metric, such as inferencing execution time or energy consumption and a target value for the metric. The augmented neural network model is characterized for various configurations and settings are determined corresponding to a variety of the performance constraints. One or more performance constraints may be provided as an input to dynamically select a configuration of the augmented neural network model. Through dynamic configuration, the augmented neural network model may adapt to real-time changes in the performance constraints. However, the trained weights for an original (before augmentation) neural network model may be used by the augmented neural network model without modification.

公开(公告)号：US09934153B2

公开(公告)日：2018-04-03

申请号：US14788593

申请日：2015-06-30

申请人： NVIDIA Corporation

发明人： Jason Lavar Clemons ,  Chih-Chi Cheng ,  Daniel Robert Johnson ,  Stephen William Keckler ,  Iuri Frosio ,  Yun-Ta Tsai

IPC分类号： G06F12/0893

CPC分类号： G06F12/0893 ,  G06F2212/454 ,  G06F2212/455

摘要： A patch memory system for accessing patches from a memory is disclosed. A patch is an abstraction that refers to a contiguous, array of data that is a subset of an N-dimensional array of data. The patch memory system includes a tile cache, and is configured to fetch data associated with a patch by determining one or more tiles associated with an N-dimensional array of data corresponding to the patch, and loading data for the one or more tiles from the memory into the tile cache. The N-dimensional array of data may be a two-dimensional (2D) digital image comprising a plurality of pixels. A patch of the 2D digital image may refer to a 2D subset of the image.