asingh9530/airdot-deployer

Airdot deployer

Airdot deployer Tool to take your ml model live right from your jupyter notebook. Before proceeding make sure you have installed docker and docker-compose.

[currently only supports local deployments]

How to install

pip install "git+https://github.com/Abhinavfreecodecamp/ml-deployer-os.git@devel#egg=airdot"

or

How to Use

# If using virtual env operator
docker network create minio-network && wget  https://github.com/Abhinavfreecodecamp/ml-deployer-os/blob/master/docker-compose.yaml && docker-compose -p airdot up

from airdot import Deployer
import pandas as pd
deployer = Deployer() 


# declare a function
df2 = pd.DataFrame(data=[[10,20],[10,40]], columns=['1', '2'])
def get_value_data(cl_idx='1'):
    return df2[cl_idx].values.tolist()

deployer.run(get_value_data) # to deploy local

How to stop

deployer.stop('get_value_data') # to stop container

How to list deployments

deployer.list_deployments() # to list all deployments

update objects

df2 = pd.DataFrame(data=[[10,20],[10,40]], columns=['1', '2'])
deployer.update_objects(('df2',df2), 'get_value_data') # to update objects like model or dataframes.