How to pass additional parameters (as a dict) to sagemeker custom inference container?

Question

Status:

• Custom container is built using the doc - https://github.com/aws/amazon-sagemaker-examples/tree/master/advanced_functionality/scikit_bring_your_own
• predict.py is coded to accommodate the custom inference script and its working well
• Using the classsagemaker.model.Model() class to pass the trained model.tar.gz and custom container image inorder to deploy the model

Challenge:

• In the same Model class there is a ENV parameter through which we can apparently send the environment variables to the custom image
• Tried passing a python dict to this , but facing difficulty to read this json dict inide the predict.py script

Somebody faced the same difficulty ?