feat: add Cohere embedding integration

[bwook00]

Description

Add Cohere Embedding provider

@Pouyanpi,
I added it along with #1304.
(It probably conflicts with 1304)

Checklist

• I've read the CONTRIBUTING guidelines.
• I've updated the documentation if applicable.
• I've added tests if applicable.
• @mentions of the person or team responsible for reviewing proposed changes.

[github-actions]

Documentation preview

https://nvidia.github.io/NeMo-Guardrails/review/pr-1305

[Copilot]

Pull Request Overview

This PR adds Cohere as a new embedding provider to the NeMo Guardrails framework, enabling users to use Cohere's embedding models for document encoding tasks.

• Implements CohereEmbeddingModel class with both synchronous and asynchronous encoding methods
• Adds comprehensive test coverage for the new Cohere integration
• Updates documentation to include Cohere in the supported embedding providers table

Reviewed Changes

Copilot reviewed 6 out of 6 changed files in this pull request and generated 1 comment.

Show a summary per file

File	Description
nemoguardrails/embeddings/providers/cohere.py	Implements the CohereEmbeddingModel class with sync/async encoding capabilities
nemoguardrails/embeddings/providers/init.py	Registers the new Cohere embedding provider in the system
tests/test_embeddings_cohere.py	Adds comprehensive test coverage for Cohere embedding functionality
tests/test_configs/with_cohere_embeddings/config.yml	Test configuration file for Cohere embeddings setup
tests/test_configs/with_cohere_embeddings/config.co	Test flow configuration for validating Cohere integration
docs/user-guides/configuration-guide.md	Updates documentation to include Cohere in supported providers table

Comments suppressed due to low confidence (1)

tests/test_embeddings_cohere.py:70

• Function name test_live_query is duplicated. This function should have a distinct name like test_live_query_sync to differentiate it from the async version on line 52.

def test_live_query(app):

[Pouyanpi]

@bwook00 Thank you for your PR 🎉

Would you please enable pre-commits and apply it? Please see contributing guidelines.

once it is enabled you can do:

poetry run pre-commit run --all-files

[codecov-commenter]

Codecov Report

❌ Patch coverage is 42.85714% with 16 lines in your changes missing coverage. Please review.
✅ Project coverage is 70.40%. Comparing base (bee719b) to head (025c4f7).
⚠️ Report is 5 commits behind head on develop.

Files with missing lines	Patch %	Lines
nemoguardrails/embeddings/providers/cohere.py	38.46%	16 Missing ⚠️

Additional details and impacted files

@@             Coverage Diff             @@
##           develop    #1305      +/-   ##
===========================================
- Coverage    70.45%   70.40%   -0.05%     
===========================================
  Files          161      162       +1     
  Lines        16214    16241      +27     
===========================================
+ Hits         11423    11434      +11     
- Misses        4791     4807      +16     

Flag	Coverage Δ
python	70.40% <42.85%> (-0.05%)	⬇️

Flags with carried forward coverage won't be shown. Click here to find out more.

Files with missing lines	Coverage Δ
nemoguardrails/embeddings/providers/__init__.py	96.42% <100.00%> (+0.13%)	⬆️
nemoguardrails/embeddings/providers/cohere.py	38.46% <38.46%> (ø)

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.
• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.

[bwook00]

@Pouyanpi Thank you for your kind review 😃
Along with this branch, I applied pre-commit too #1304

[bwook00]

Hi @Pouyanpi ,

Just wanted to gently ask if there are any plans to merge this PR.

[Pouyanpi]

Hi @bwook00, I apologize for the significant delay and truly appreciate your patience on this PR.

What blocked the merge:

The blocker wasn't issues with your code. it's that we don't have proper test infrastructure in place for external API providers. I've been meaning to implement a mock-based testing framework for this, but haven't found the time to do it properly.

Currently, your tests follow the same pattern as our OpenAI tests (requiring LIVE_TEST mode with valid API credentials). While this works for manual testing, it means:

• the tests don't run in CI/CD
• we can't test error conditions or edge cases without actually hitting Cohere's API
• coverage is limited to the "happy path" with known models

Why the coverage is 42.85%:

Looking at your implementation, the 16 missing lines are likely:

1. import error handling (lines 58-62), when cohere package isn't installed
2. unknown model handling (lines 79-80), the fallback when a model isn't in the predefined dictionary
3. different input_type values, your code supports various input types but tests only use the default
4. client initialization with different kwargs, not tested with various parameters
5. API error conditions, network failures, invalid API keys, rate limits, etc.

I am going to push those tests directly to this branch (or open a PR and target to your branc) and then you can take over from there. What do you think?

[bwook00]

Hi @Pouyanpi ,
First of all, thank you for your review!

I am going to push those tests directly to this branch (or open a PR and target to your branc) and then you can take over from there. What do you think?

Sounds good 🙂. Please let me know once you've pushed the tests to this branch and where I should take over from. Alternatively, I'm also happy to add those tests myself. Just let me know whichever way is more convenient for you!

So, just to confirm my understanding of the next steps:

1. Wait for test: add mock embedding provider tests #1446 to be merged.
2. Take over this PR to complete the tests.
3. Then, finish the modifications on feat: add Google embedding integration #1304 and wait for its merge.

Could you please confirm if this is correct? :)

[Pouyanpi]

Thank you @bwook00 for your flexibility and active contribution.

So, just to confirm my understanding of the next steps:

1. Wait for test: add mock embedding provider tests #1446 to be merged.
2. Take over this PR to complete the tests.

Yes, this is correct. You just need to remove the part that skips the Cohere tests, which was introduced in #1446.
You’re also welcome to add more tests if you think they’re needed or if you notice any inconsistencies between the actual API and our mocks. Otherwise, I think this is good to merge.

3. Then, finish the modifications on feat: add Google embedding integration #1304 and wait for its merge.

You can start working on that PR now. It’s blocked by #1446, not #1305. We’ll be merging #1446 by the end of the day.

[bwook00]

@Pouyanpi , thank you for your kind reply!

I’ve already started #1304. Could you let me know once #1446 is merged?

[Pouyanpi]

@bwook00 #1446 is merged to develop. Please rebase your branch onto develop and then remove following lines from the test_embeddings_providers_mock.py and push.

try:
    import nemoguardrails.embeddings.providers.cohere

    COHERE_AVAILABLE = True
except (ImportError, ModuleNotFoundError):
    COHERE_AVAILABLE = False


@pytest.mark.skipif(
    not COHERE_AVAILABLE, reason="Cohere provider not available in this branch"
)