feat(fastembed): allow passing model_kwargs to dense embedders

[RahilOp]

Related Issues

• fixes Allow passing provider option to FastEmbed to allow embedding on the GPU #1575

Proposed Changes:

FastembedTextEmbedder and FastembedDocumentEmbedder did not expose a way to forward extra arguments to the underlying fastembed TextEmbedding model. As a result, users could not select an ONNX execution provider (e.g. CUDAExecutionProvider) to run embeddings on GPU, which is the standard way to enable GPU support in fastembed.

This adds a model_kwargs parameter to both dense embedders and the dense embedding backend, forwarding the kwargs to TextEmbedding(...). This mirrors the existing model_kwargs support already present in the sparse embedders (FastembedSparseTextEmbedder / FastembedSparseDocumentEmbedder), so the design and serialization are consistent across the integration.

Usage:

embedder = FastembedTextEmbedder(
    model="BAAI/bge-small-en-v1.5",
    model_kwargs={"providers": ["CUDAExecutionProvider"]},
)

Note: running on GPU also requires installing fastembed-gpu in the environment (as per the fastembed docs); this PR only exposes the passthrough so the option becomes usable, without changing the integration's dependencies.

How did you test it?

Unit tests (hatch run test:unit), following the existing sparse-embedder test patterns:

• backend forwards model_kwargs to TextEmbedding
• model_kwargs is stored on both embedders and round-trips through to_dict
• warm_up passes model_kwargs to the backend factory

AI tooling assisted with parts of this change.

Notes for the reviewer

The dense backend cache key now also includes local_files_only and model_kwargs, matching the sparse backend's key so backends differing only in those args aren't incorrectly shared.

Checklist

• I have read the contributors guidelines and the code of conduct
• I added unit tests and updated the docstrings
• I've used a conventional commit type for my PR title: feat:

[github-actions]

Heads-up for maintainers

This PR is from a fork and touches integrations whose integration tests require API keys.
Those tests are skipped in CI because fork PRs don't have access to repo secrets for security reasons.

Affected integrations:

• fastembed

Please run the integration tests locally (hatch run test:integration inside each folder) before approving.

[github-actions]

Coverage report (fastembed)

Click to see where and how coverage changed

File	Statements	Missing	Coverage	Coverage (new stmts)	Lines missing
integrations/fastembed/src/haystack_integrations/components/embedders/fastembed
fastembed_document_embedder.py
fastembed_text_embedder.py
integrations/fastembed/src/haystack_integrations/components/embedders/fastembed/embedding_backend
fastembed_backend.py
Project Total

This report was generated by python-coverage-comment-action