Is there a way to deploy an AutoGluon pre-built model in AWS?

Question

I have generated a model for text Prediction using AutoGluon running in a SageMaker notebook instance. I already have a model folder, containing the following files:

hf_text folder;
assets.json
config.yaml
data_processor.pkl
df_preprocessor.pkl
DS_Store
model.ckpt

The model is working well but I am facing problems to deploy it to be called as an inference endpoint. All the documentation found is using TabularPredictor, but I need help for TextPredictor

Is the answer understandable and good for you? Please mark it as an answer if it is good. — zhrist, Oct 22 '22 at 18:46

score 0 · Answer 1 · answered Sep 28 '22 at 13:04

Have you created a Model object? This means that you would link a model file(s3 location) with DLC(deep learning container) which will enable you to deploy it to an Endpoint, what was your question.

The workflow is : Create Model -> Create Endpoint Configuration -> Create endpoint.

Here is an more detailed example with code, so you can figure out the rest.

Is there a way to deploy an AutoGluon pre-built model in AWS?

1 Answers1