Methods, systems, articles of manufacture, and apparatus to optimize layers of a machine learning model for a target hardware platform
AssigneeIntel Corporation
InventorAmit Bleiweiss
Life of this patent
Estimating the term…
Abstract
Methods, apparatus, systems, and articles of manufacture are disclosed that optimize layers of a machine learning model for a target hardware platform. An example apparatus includes a communication processor to obtain information specific to the target hardware platform (THP) on which to execute the machine learning model; a layer generation controller to generate layers of the machine learning model based on the information specific to the THP; and a deployment controller to, in response to the machine learning model satisfying a threshold error metric, deploy the machine learning model to the THP.
Claims21
An apparatus comprising:
- interface circuitry to obtain information specific to a target hardware platform (THP) on which a machine learning model is to be executed;
- instructions;
- and
- at least one processor circuit to be programmed by the instructions to:
- select a first hidden state and a second hidden state for a first layer of the machine learning model;
- select a first operation to apply to the first hidden state based on the information specific to the THP;
- select a second operation to apply to the second hidden state based on the information specific to the THP;
- recursively add one or more additional layers to the machine learning model based on the information specific to the THP until the machine learning model includes a threshold number of layers;
- and
- cause the interface circuitry to deploy the machine learning model with the threshold number of layers to the THP.
A non-transitory computer-readable medium comprising instructions to cause at least one processor to at least:
- obtain information specific to a target hardware platform (THP) on which a machine learning model is to be executed;
- select a first hidden state and a second hidden state for a first layer of the machine learning model;
- select a first operation to apply to the first hidden state based on information specific to the THP;
- select a second operation to apply to the second hidden state based on the information specific to the THP;
- recursively add one or more additional layers to the machine learning model based on the information specific to the THP until the machine learning model includes a threshold number of layers;
- and
- cause interface circuitry to deploy the machine learning model with the threshold number of layers to the THP.
An apparatus comprising:
- means for processing communications to obtain information specific to a target hardware platform (THP) on which a machine learning model is to be executed;
- means for selecting hidden states to select a first hidden state and a second hidden state for a first layer of the machine learning model;
- means for selecting operations to:
- select a first operation to apply to the first hidden state based on the information specific to the THP;
- and
- select a second operation to apply to the second hidden state based on the information specific to the THP;
- means for generating layers of the machine learning model to recursively add one or more additional layers to the machine learning model based on the information specific to the THP until the machine learning model includes a threshold number of layers;
- and
- means for deploying machine learning models to cause the means for processing communications to deploy the machine learning model with the threshold number of layers to the THP.
A server to distribute first instructions on a network, the server comprising:
- at least one storage device including second instructions;
- and
- at least one processor to execute the first instructions and the second instructions to transmit the first instructions over the network, the first instructions to cause at least one device to:
- obtain information specific to a target hardware platform (THP) on which a machine learning model is to be executed;
- select a first hidden state and a second hidden state for a first layer of the machine learning model;
- select a first operation to apply to the first hidden state based on the information specific to the THP;
- select a second operation to apply to the second hidden state based on the information specific to the THP;
- recursively add one or more additional layers to the machine learning model based on information specific to the THP until the machine learning model includes a threshold number of layers;
- and
- deploy the machine learning model with the threshold number of layers to the THP.
Dependencies are inferred from the claim text.