How to fetch a model and make predictions from the Model Registry

Question

Hello! I have a model in the Sagemaker Studio Model Registry. Is there a way to do something like the following?

```
sm = boto3.client('sagemaker')
model = sm.fetch_model('model_group_name')
preds = model.predict(X)
```

I'm used to using MLFlow, where you can do 
```
import mlflow.sklearn

sk_model = mlflow.sklearn.load_model("runs:/96771d893a5e46159d9f3b49bf9013e2/sk_models")

# use Pandas DataFrame to make predictions
pandas_df = ...
predictions = sk_model.predict(pandas_df)

```

I can't find an equivalent of the `sm.fetch_model()` function. I'm looking for something along the lines of a MLFlow equivalent. I want to do some predictions without having a model deployed.

Answer

SageMaker Model Registry can be used to catalog and manage different model versions. model_group_name can't be used directly from Sagemaker Model Registry. First you need to deploy the model. After deploying the model, Endpoint would be created. Then you can directly invoke the Endpoint API.

Generally at first Model You need to to create a Model group[1] << Register a Model Version using Sagemaker Model registry [2] << Update the Approval Status of a Model to approved[3] << Deploy the Model for registry[4] << Invoke the Endpoint API.

Kindly go through the below example of deploying the model and then invoking it via Endpoint API
https://github.com/aws/amazon-sagemaker-examples/blob/main/serverless-inference/serverless-model-registry.ipynb

[1] https://docs.aws.amazon.com/sagemaker/latest/dg/model-registry-model-group.html
[2] https://docs.aws.amazon.com/sagemaker/latest/dg/model-registry-version.html
[3] https://docs.aws.amazon.com/sagemaker/latest/dg/model-registry-approve.html
[4] https://docs.aws.amazon.com/sagemaker/latest/dg/model-registry-deploy.html

How to fetch a model and make predictions from the Model Registry

Contenuto pertinente