seldon-mab

Chart to deploy a multi-armed bandits router over two Seldon Core v1 deployments, so that traffic is sent to the best performing model. You will need to utilize both the predict and send_feedback API methods.

Usage

To use this chart, you will first need to add the seldonio Helm repo:

helm repo add seldonio https://storage.googleapis.com/seldon-charts
helm repo update

Once that’s done, you should then be able to use the inference graph template as:

helm template $MY_MODEL_NAME seldonio/seldon-mab --namespace $MODELS_NAMESPACE

Note that you can also deploy the inference graph directly to your cluster using:

helm install $MY_MODEL_NAME seldonio/seldon-mab --namespace $MODELS_NAMESPACE

Homepage: https://github.com/SeldonIO/seldon-core

Source Code

• https://github.com/SeldonIO/seldon-core

• https://github.com/SeldonIO/seldon-core/tree/master/helm-charts/seldon-mab

Values

Key	Type	Default	Description
engine.env.SELDON_LOG_MESSAGES_EXTERNALLY	bool	false
engine.env.SELDON_LOG_MESSAGE_TYPE	string	"seldon.message.pair"
engine.env.SELDON_LOG_REQUESTS	bool	false
engine.env.SELDON_LOG_RESPONSES	bool	false
engine.resources.requests.cpu	string	"0.1"
mab.branches	int	2
mab.epsilon	float	0.2
mab.image.name	string	"seldonio/mab_epsilon_greedy"
mab.image.version	string	"1.17.0-dev"
mab.name	string	"eg-router"
mab.verbose	int	1
modela.image.name	string	"seldonio/mock_classifier"
modela.image.version	string	"1.17.0-dev"
modela.name	string	"classifier-1"
modelb.image.name	string	"seldonio/mock_classifier"
modelb.image.version	string	"1.17.0-dev"
modelb.name	string	"classifier-2"
predictor.name	string	"default"
predictorLabels.fluentd	string	"true"
predictorLabels.version	string	"1.17.0-dev"
replicas	int	1
sdepLabels.app	string	"seldon"