Feature Overview 

• Productize the Arctic embedding model (https://huggingface.co/Snowflake/snowflake-arctic-embed-l-v2.0 ) in RHEL AI to enable subset selection. 

Goals

• Unblock SDG team to support subset selection capability
• Enable end users to use subset selection to improve experience and performance when they have a large dataset size 

Requirements:

• Investigate performance of Arctic large embedding model on CPU vs GPU
• Default Arctic embedding model on GPU for RHEL AI with option to run it on CPU (e.g. IBM InstructLab service)

Done - Acceptance Criteria:

• Get approval from legal department
• Build and publish Arctic embedding model artifact for RHEL AI

Use Cases - i.e. User Experience & Workflow:

• Subset Selection capability built by Red Hat Innovation team requires this new embedding model

Documentation Considerations:

• Transparency -> end user is aware subset selection uses Arctic Emebdding model.
• Include instructions to pull Arctic model from Red Hat registry to RHEL AI

Out of Scope:

The embedding model is not exposed to user for other use cases in RHEL AI, it is an implementation detail for the subset selection capability.

Background and Strategic Fit (Initial completion while in Refinement status):
Provide any additional context is needed to frame the feature.
<your text here>

Customer Considerations {}{}(Initial completion while in Refinement status):
Provide any additional customer-specific considerations that must be made when designing and delivering the Feature.
<your text here>