Back to results
US 2021/0174202 A1Published application
Method and apparatus with model optimization, and accelerator system
AssigneeSamsung Electronics Co., Ltd.
InventorJae-Ki HONG
Life of this patent
Estimating the term…
Abstract
A method with model optimization includes: determining a graph representing operations performed in a target model; determining an attribute of input data of the target model; determining a predicted performance of the target model based on a behavior pattern of hardware that executes the target model; and optimizing the operations performed in the target model based on the graph, the attribute of the input data, and the predicted performance of the target model.
Claims22
A method with model optimization, comprising:
- determining a graph representing operations performed in a target model;
- determining an attribute of input data of the target model;
- determining a predicted performance of the target model based on a behavior pattern of hardware that executes the target model;
- and
- optimizing the operations performed in the target model based on the graph, the attribute of the input data, and the predicted performance of the target model.
2 more elements
An apparatus with model optimization, comprising:
- at least one processor configured to:
- determine a graph representing operations performed in a target model;
- determine an attribute of input data of the target model;
- determine a predicted performance of the target model based on a behavior pattern of hardware that executes the target model;
- and
- optimize the operations performed in the target model based on the graph, the attribute of the input data, and the predicted performance of the target model.
3 more elements
An accelerator system, comprising:
- a model optimization apparatus configured to optimize operations performed in a target model based on a graph representing the operations, an attribute of input data of the target model, and a predicted performance of the target model determined based on a behavior pattern of an accelerator in which the target model is executed;
- and
- an accelerator configured to infer input data using a target model to which the optimized operations are applied.
Dependencies are inferred from the claim text.