Sagemaker deployment template.

#1
by fprolog - opened

Is the sagemaker snippet ready for deployment , if so, how is the serializer expecting the input ?

I would also like an example of the predictor's input

I would also like an example of the predictor's input

In case you still need it , I made it work using directly the pytorch serving libraries of AWS https://sagemaker.readthedocs.io/en/stable/frameworks/pytorch/using_pytorch.html#deploy-pytorch-models

Thanks. I am familiar with the sagemaker model pipeline. My problem is due to the Image + text nature of this model. When I invoke "response = predictor.predict(data)" I would like an example of the data object containing an image and a prompt for the model.

fprolog changed discussion status to closed

Sign up or log in to comment