Skip to content

2.0 Lite Support, Inference Adapter extensibility, SM Inference Adapter support#52

Merged
VikramShenoy97 merged 1 commit intomainfrom
nova-2-lite-support
Mar 13, 2026
Merged

2.0 Lite Support, Inference Adapter extensibility, SM Inference Adapter support#52
VikramShenoy97 merged 1 commit intomainfrom
nova-2-lite-support

Conversation

@VikramShenoy97
Copy link
Contributor

Issue #, if available:

Description of changes:

  • Added Nova 2.0 Lite model support, replacing Nova Premier references with Nova 2.0 Lite across source, tests, and documentation
  • Introduced inference adapter extensibility: refactored InferenceAdapter into separate BedrockInferenceAdapter and SageMakerInferenceAdapter implementations, with a DSPy-compatible adapter layer (dspy_compatible.py)
  • Added SageMaker inference adapter support, enabling prompt optimization against SageMaker-hosted endpoints with OpenAI-compatible message format
  • NovaPromptOptimizer now supports separate inference adapters for meta-prompting and task optimization phases, allowing Bedrock for meta-prompting while optimizing a SageMaker endpoint
  • Added new lite-2 optimization mode targeting the Nova 2.0 Lite task model
  • Updated sample notebooks with refreshed optimization outputs
  • Bumped version to 1.1.0
  • Added unit and integration tests for all new adapters (195 tests passing)

By submitting this pull request, I confirm that you can use, modify, copy, and redistribute this contribution, under the terms of your choice.

@@ -10,20 +10,23 @@
},
{
"cell_type": "code",
"execution_count": null,
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I would favor unexecuted notebooks as a sample. Is there a reason why we have the outputs of an execution populated?

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

This was the case previously as well. It shows sample Prompt Optimization and Gains on the sample data

- Cross-region support
- Independent rate limiting per adapter

For more details, see the [Separate Inference Adapters Guide](docs/SeparateInferenceAdapters.md).
Copy link

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I do not see this doc in this PR or main... did we miss it?

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Thanks for catching that. I'll double check on those in another PR

print("✗ Endpoint connection failed")
```

For a complete guide, see the [SageMaker Quick Start](docs/QUICK_START_SAGEMAKER.md).
Copy link

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Same as above:I do not see this doc in this PR or main... did we miss it?

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Thanks for catching that. I'll double check on those in another PR

@VikramShenoy97 VikramShenoy97 merged commit ae09d72 into main Mar 13, 2026
5 checks passed
@VikramShenoy97 VikramShenoy97 deleted the nova-2-lite-support branch March 13, 2026 18:07
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

4 participants