DO NOT USE - This is a sample program
Project description
Servz
Machine Learning Model Serving
This library is a pre-alpha right now
The intent of this package is to provide a machine learning deplpoyment layer for model predictions.
Features
- pipeline driven
- deployment via task runner
- deployment via flask or other endpoint
- MLFlow based deployment
- Seldon deployment (in progress)
Pipeline formats
Pipeline Composer
Appends all the pipelines from the yaml file(s)
"pipelines": self._pipelines
return [self.__build_flow(pipeline) for pipeline in flows['pipelines']]
self._pipelines.append(pipeline)
Artifact Builder
"artifact": self._workflows
_res = self._build(pipelines=kwargs.get('packager'))
results = [self._compile_workflow(pipe) for pipe in pipelines]
self._workflows = pipelines
for task in pipe['workflow']:
do stuff
success = self.__build_prefect_flow()
return {
'success': success,
}
Endpoint Appender
Packager
Package Publisher
structure
/core - core files for servz
/orchestration_artifact_builder - artifact packager
/orchestration_artifact_deployer - deployment runner
/packager - manifest packager
/pipeline - loading and validation of serving pipeline
/server_templates - artifacts for artifact builder to construct endpoints
/tests - unit tests and e2e tests
```
Project details
Release history Release notifications | RSS feed
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
Source Distribution
servz-0.0.0.38.tar.gz
(13.1 kB
view hashes)