On-Boarding User Guide¶
This is the users guide to Onboarding.
1: introduction - What is Onboarding?¶
Acumos is intended to enable the use of a wide range of tools and technologies in the development of machine learning models including support for both open sourced and proprietary toolkits.
The goal of Onboarding is to provide an ingestion interface, by web or CLI(command line interface) for various types of models and to create required artifacts and identifiers to enter the Acumos platform.
- Legacy models
You can on-board models developped in Java 8 or 9, Python>=3.6, <3.7, R>=3.4.4 and sourced from toolkits such as Scikit, TensorFlow, H2O, and R. You can choose to create or not the microservice at the end of the on-boarding process. If user choose to not create the microservice at the end of on-boarding he can create it later.
In short, our goals for these kinds of models are to generate or provide all the necessary materials required to use these kinds of models with the others components of Acumos like:
- Tosca file for Design studio
- Represent model I/O such for microservice generation
- SolutionID for CDS
- Licence file for licensing management
- Since Boreas, we are able to on-board news kinds of model like :
- model in onnx format : model.onnx
- model in pfa format : model.pfa
- Dockerized model : model dockerised by Data scientist himself
- Docker model URI : URI of Dockerized model stored in external repo like Docker-Hub for example
For this new kinds of model, Micro service generation and Design studio capabilities are not available.
- Acumos capabilities by models type
This table sum-up all the Acumos capabilities available for each kinds of model
Model Micro-service generation Design studio Market place on-board with license onboarding R model Available Available Available Available WEB and CLI Pyhton model Available Available Available Available WEB and CLI Java model Available Available Available Available WEB and CLI ONNX model Not available Not available Available Available WEB only PFA model Not available Not available Available Available WEB only Dockerized model Not applicable Not available Available Not available WEB only URI model Not applicable Not applicable Available Available WEB only C++ model Available Not applicable Available Available WEB only
3: Onboarding models built in R, Java or python language¶
For these three different languages it exists three on-boarding acumos clients to use. The client will build a model bundle composed of differents files requested by Acumos. Once the model bundle is created you can choose to on-board it by CLI or WEB. By CLI on-boarding you will have to use a push() function that belongs to the Acumos on-boardign client, if you choose Web on-boarding you have to use the “ON-BOARDING MODEL” panel of you own Acumos instance. Whatever the kind of on-boarding, by WEB or by CLI, you can choose to trigger or not the launch of the micro-service generation at the end of the on-boarding process.
Whatever the acumos client you used, you will be prompted to provide your credentials in this way : “your_acumos_login”:”your_api_token”. Your api token can be retrieved in your own acumos instance, after authentication, in your acumos settings. After a successful CLI onboarding with Micro-service creation, the message response will display the acumos Docker URI that can be used to load the Acumos docker image model in your local Docker registry.
Please refer to the following user guide :
4: Onboarding ONNX and PFA models¶
Onboard ONNX and PFA model consists only of an upload of the model as there is no micro-service creation for the moment.
Please refer to the following user guide
5: On-Boarding docker image model or docker URI model¶
You can create models in the language of your choice then dockerize your models yourelves and onboard these dockerized models or dockerized model URIs. Of course for these kinds of models the microservice generation process is never used.
Please refer to the following user-guide
On-Boarding docker image model or docker URI model user guide
5: On-Boarding model with a license file¶
You can on-board your model with a license (Except for dockerized models as we assume that modelers will embed their licence in their Docker image). During Web on-boarding you can choose “Upload” a license file, “Select licence Profil” or “Create New”, please refer to licensing documentation for the details. With CLI on-boarding only the “Upload” of a license file is possible alongside the model model bundle.
Whatever the case, CLI or WEB on-boarding :
- If the license file extension is not ‘json’ the license on-boarding will not be possible.
- During the on-boarding, your license file will be renamed and you will see your license file as “license-1.0.0.json” in the artifact table.
- If you on-board a new version of your model through the portal, the license number revision will be increased like that “license-1.0.1.json” to follow the model number revision.
Whatever the kinds of models :
- New solution is created in common database for a new model.
- Existing solution is updated with, a new revision. Revision is updated with artefact details and those artefacts are uploaded to nexus maven repository.
