Does the pre and post-processing need to be incorporate in SageMaker?

0

Hi All,

Greetings!!

Please address my below queries.

  1. Does the pre and post-processing need to be incorporate in SM?
  2. Isn't SM supposed to be used for inference only?

Why I'm asking these questions because I am using PyTorch model server for pre-processing, predictions and post-processing for NER use cases.

  1. We use utokenize for creating tokens which supports >= Python 3.8 but PyTorch model server supports <= 3.6, what would be the solution? How to install Python 3.8 in PyTorch model server?

  2. Don't we have basic word piece tokenization as pre-processing in SM?

Thanks, Vinayak

asked 10 months ago87 views
3 Answers
0

Hi Vinayak

I just want to make sure I understand the context: Are you referring to pre and postprocessing for an inference request or for training a model? I'm asking because SageMaker offers both:

Could you help clarify the question?

Thanks Heiko

Heiko
answered 10 months ago
0

Hello @Heiko,

I am talking about inference pipeline only.

Don't we have pre and post-processing in-built feature for NER in SM PyTorch or HuggingFace model servers?

We thought that pre and post-processing taken care by SM and we just need to bring our fine-tuned model on our dataset.

Thanks, Vinayak

answered 10 months ago
0

Hi Vinayak - apologies for the late response, I was OOTO.

Yes, you are correct, pre- and post-processing for Huggingface inference is "built in" via the SageMaker Hugging Face Inference Toolkit. The documentation on the Github repo shows how this is being done.

Hope that helps!

Thanks Heiko

Heiko
answered 9 months ago

You are not logged in. Log in to post an answer.

A good answer clearly answers the question and provides constructive feedback and encourages professional growth in the question asker.

Guidelines for Answering Questions