CODEWITHSUNDEEP
pythonamazon-web-servicesmachine-learningamazon-sagemakerxgboost
Nikhil Kumar AV
Nikhil Kumar AV

Reputation: 11

How to load Sagemaker XGBoost model which was trained using (version = 'latest') framework?

There is an existing xgboost model in the pipeline that was created using this container

sagemaker.image_uris.retrieve('xgboost', sagemaker.Session().boto_region_name, version='latest')

output:

'{accountid}.dkr.ecr.us-east-1.amazonaws.com/xgboost:latest'

I extracted the model.tar.gz from the model artifact and loaded the xgboost-model file
but it gave this error

XGBoostError: basic_string::resize

I ran a shell script that loads the model using all the versions of XGBoost available but nothing worked.
I just want to check the feature importance using model.get_score.

Upvotes: 0

Views: 25

Answers (1)

Tomonori Shimomura
Tomonori Shimomura

Reputation: 267

As a first step to investigate this error, please check if this error reproduces in your local environment by installing xgboost locally.

If this error reproduces also locally, you can eliminate amazon-web-services / amazon-sagemaker tags from this post.

It is also recommended to add some information how you are calling the xgboost APIs to load the model, and how your model files are structured under the extracted model directory, as the error may be happening because of how the model is structured, and how to load it.

Upvotes: 0

Related Questions