Model ML registry and model serving
AssigneeDatabricks, Inc.
InventorsAaron Daniel Davidson, Tomas Nykodym, Clemens Mewald
Life of this patent
Estimating the term…
Abstract
A system includes an interface, a processor, and a memory. The interface is configured to receive a version of a model from a model registry. The processor is configured to store the version of the model, start a process running the version of the model, and update a proxy with version information associated with the version of the model, wherein the updated proxy indicates to redirect an indication to invoke the version of the model to the process. The memory is coupled to the processor and configured to provide the processor with instructions.
Claims24
A computer system, comprising:
- a computer processor;
- and
- a memory comprising stored instructions that when executed by the computer processor causes the computer system to:
- an interface configured to:
- provide access to a model registry, wherein the model registry is a database for storing one or more versions of a machine-learned model that is trained via a machine-learning process;
- receive an indication to update a version of a machine-learned model in the model registry;
- receive an updated version of the machine-learned model from a user device;
- store the updated version of the machine-learned model in the model registry;
- update, via the computer system, a proxy with version information associated with the updated version of the machine-learned model to generate an updated proxy, wherein the updated proxy stores request routing information associated with a process running the updated version of the machine-learned model indicates;
- receive a request to perform inference using the updated version of the machine-learned model, the request including input data for input into the machine-learned model;
- and
- perform the request by routing the request to the process running the updated version of the machine-learned model based on the request routing information stored in the updated proxy.
A method, comprising:
- providing access to a model registry, wherein the model registry is a database for storing one or more versions of a machine-learned model that is trained via a machine-learning process;
- receiving an indication to update a version of a machine-learned model in the model registry;
- receiving an updated version of the machine-learned model from a user device;
- storing the updated version of the machine-learned model in the model registry;
- updating, via a computer system, a proxy with version information associated with the updated version of the model to generate an updated proxy, wherein the updated proxy stores request routing information associated with a process running the updated version of the machine-learned model;
- receiving a request to perform inference using the updated version of the machine-learned model, the request including input data for input into the machine-learned model;
- and
- performing the request by routing the request to the process running the updated version of the machine-learned model based on the request routing information stored in the updated proxy.
A non-transitory computer-readable medium storing instructions that, when executed by a processor, cause the processor to:
- provide access to a model registry, wherein the model registry is a database for storing one or more versions of a machine-learned model that is trained via a machine-learning process;
- receive an updated version of the machine-learned model from a user device;
- store the updated version of the machine-learned model in the model registry;
- update, via a computer system, a proxy with version information associated with the updated version of the model to generate an updated proxy, wherein the updated proxy stores request routing information associated with a process running the updated version of the machine-learned model;
- receive a request to perform inference using the updated version of the machine-learned model, the request including input data for input into the machine-learned model;
- and
- perform the request by routing the request to the process running the updated version of the machine-learned model based on the request routing information stored in the updated proxy.
Dependencies are inferred from the claim text.