Standardized model packaging and deployment
Standardized model packaging and deployment, including: generating a model package comprising: model definition data for a model; function code facilitating execution of the model; and at least one interface for at least one operating system.
1. A method implemented by a computing device for standardized model packaging and deployment, the method comprising:
generating a model package comprising:
for a model, model definition data that includes a binding to a computer programming language, wherein the model is a trained machine learning model to generate prediction data, wherein the binding enables one or more functions written in the computer programming language to execute one or more portions of the model;
function code that facilitates execution of the model via the binding;
at least one interface for at least one operating system; and
wherein one or more of the model definition data or the function code access one or more runtime dependencies excluded from the model package.
2. The method of claim 1 , wherein the model definition data is encoded in a model definition language (MDL).
3. The method of claim 2 , wherein the computer programming language encodes the function code.
4. The method of claim 3 , wherein the model definition data comprises one or more other bindings to one or more other programming languages.
5. The method of claim 1 , wherein the at least one interface comprises an Application Binary Interface (ABI).
6. The method of claim 1 , wherein the at least one interface comprises a Foreign Function Interface (FFI).
7. The method of claim 1 , further comprising deploying the model package for execution of the model.
8. The method of claim 7 , further comprising executing the model via the model package.
9. The method of claim 1 , further comprising:
training the model based on input training data; and
wherein generating the model package comprises exporting, after training the model, the model as the model definition data.
10. An apparatus for standardized model packaging and deployment, the apparatus comprising a computer processor, a computer memory operatively coupled to the computer processor, the computer memory having disposed within it computer program instructions that, when executed by the computer processor, cause the apparatus to perform steps comprising:
generating a model package comprising:
for a model, model definition data that includes a binding to a computer programming language, wherein the model is a trained machine learning model to generate prediction data, wherein the binding enables one or more functions written in the computer programming language to execute one or more portions of the model;
function code that facilitates execution of the model via the binding;
at least one interface for at least one operating system; and
wherein one or more of the model definition data or the function code access one or more runtime dependencies excluded from the model package.
11. The apparatus of claim 10 , wherein the model definition data is encoded in a model definition language (MDL).
12. The apparatus of claim 11 , wherein the computer programming language encodes the function code.
13. The apparatus of claim 12 , wherein the model definition data comprises one or more other bindings to one or more other programming languages.
14. The apparatus of claim 10 , wherein the at least one interface comprises an Application Binary Interface (ABI).
15. The apparatus of claim 10 , wherein the at least one interface comprises a Foreign Function Interface (FFI).
16. The apparatus of claim 10 , wherein the steps further comprise deploying the model package for execution of the model.
17. A computer program product disposed upon a non-transitory computer readable medium, the computer program product comprising computer program instructions for standardized model packaging and deployment that, when executed, cause a computer system to carry out the steps of:
generating a model package comprising:
for a model, model definition data that includes a binding to a computer programming language, wherein the model is a trained machine learning model to generate prediction data, wherein the binding enables one or more functions written in the computer programming language to execute one or more portions of the model;
function code that facilitates execution of the model via the binding;
at least one interface for at least one operating system; and
wherein one or more of the model definition data or the function code access one or more runtime dependencies excluded from the model package.
18. The computer program product of claim 17 , wherein the model definition data is encoded in a model definition language (MDL).