Tutorial for ml inference with cohere rerank model

[brianf-aws]

Description

Currently there are tutorials that exist using the rerank pipeline. However its possible to do the same operation with the ML_inference processor with the by_field rerank.

Check List

• New functionality includes testing.
• New functionality has been documented.
• API changes companion pull request created.
• Commits are signed per the DCO using --signoff.
• Public documentation issue/PR created.

By submitting this pull request, I confirm that my contribution is made under the terms of the Apache 2.0 license.
For more information on following Developer Certificate of Origin and signing off your commits, please check here.

[mingshl]

Hi @brianf-aws, I am thinking we should have a new folder call ml_inference that can host all use cases combing using ml_inference processors.

[dhrubo-os]

Hi @brianf-aws, I am thinking we should have a new folder call ml_inference that can host all use cases combing using ml_inference processors.

And in that folder we can have sub folder for managed service vs opensource so that the blueprint doesn't seem too long.

@brianf-aws Could you also take update from main. Seems like the integ test doesn't run. I know for blueprint integ test doesn't matter.

I just wanted to check if codecov is activate properly or not.

[mingshl]

Hi @brianf-aws, I am thinking we should have a new folder call ml_inference that can host all use cases combing using ml_inference processors.

maybe the file path can be docs/tutorials/ml_Inference/rerank/ml_Inference_with_Cohere_Rerank_model.md

[brianf-aws]

Sorry @mingshl and @dhrubo-os I don't think that would benefit the community that much currently in the directory folder we have 4 documents that mention the ml_inference processor.

### I'm in /ml-commons/docs
 docs % echo pwd  | grep -i -r -l --include="*.md" -e "ml inference processor" -e "ML processor" -e "ml inference ingest processor" -e "ml_inference" .
./tutorials/rerank/ml_Inference_with_Cohere_Rerank_model.md
./tutorials/semantic_search/asymmetric_embedding_model.md
./remote_inference_blueprints/amazon_comprehend_connector_blueprint.md
./remote_inference_blueprints/amazon_textract_connector_blueprint.md

We will make it harder for users to find answers. Consider a community user that wants to rerank documents using our project. they probably will be surprised to find out that they need to look at a directory called ml_inference/rerank.

I think perhaps the best case would be to create a file pointing to these tutorials that show how significant the ML inference processor is.

[mingshl]

./remote_inference_blueprints/amazon_comprehend_connector_blueprint.md
./remote_inference_blueprints/amazon_textract_connector_blueprint.md

In the blueprint that use ml_inference processor that was just to demo one of the usage of the model. That's totally fine.

but these two can belong to ml_inference folder

./tutorials/rerank/ml_Inference_with_Cohere_Rerank_model.md
./tutorials/semantic_search/asymmetric_embedding_model.md

what do you think? @ylwu-amzn @dhrubo-os

[dhrubo-os]

./remote_inference_blueprints/amazon_comprehend_connector_blueprint.md
./remote_inference_blueprints/amazon_textract_connector_blueprint.md

In the blueprint that use ml_inference processor that was just to demo one of the usage of the model. That's totally fine.

but these two can belong to ml_inference folder

./tutorials/rerank/ml_Inference_with_Cohere_Rerank_model.md ./tutorials/semantic_search/asymmetric_embedding_model.md

what do you think? @ylwu-amzn @dhrubo-os

I see this as a similar issue when we write a long class and then later we start having confusion where to put this class as this serves multiple responsibilities.

Take ./tutorials/semantic_search/asymmetric_embedding_model.md as an example.

Currently, asymmetric_embedding_model.md tackles both generating asymmetric embeddings and performing semantic search with the ML-Inference processor. To improve clarity and maintainability, I suggest splitting it into two focused tutorials: one for generating embeddings and another for applying them in semantic search. This would make each tutorial easier to follow, reusable in other contexts, and simpler to maintain.

[brianf-aws]

Hey, @mingshl I refactored to introduce a ml_inference folder in this commit 3986b94. Apologies for the late change

[mingshl]

nice! keep this going!!!