Support for local/in-process embedding libraries
Nobody has claimed this yet.
- Dominant language
- Rust
- Stars
- 83
- Forks
- 4
- PR merge metrics
- No merged PRs in 30d
Description
For offline or cost-free embeddings, local libraries like FastEmbed
or sentence-transformers are useful. These run in-process without HTTP.
Possible API
client = Client()
response = client.embed(
model="fastembed/BAAI/bge-small-en-v1.5",
inputs=["Hello world"]
)
This would need a different provider type (no HTTP client) and could be an optional
dependency to keep the core package light.
Happy to discuss if this fits the project scope. It's a different paradigm from
API-based providers but would be quite handy for local development and testing.
Thanks for building catsu!
Contributor guide
First steps
- Read the whole issue, then the project's contributing guide.
- Comment on the issue to say you are picking it up — it saves two people doing the same work.
- Fork the repository and make your change on a branch.
- Open a pull request that references the issue number.
Research direction
Start by reviewing the existing Client API and provider architecture, then compare the proposed FastEmbed and sentence-transformers usage with the current HTTP-based design. Determine how an in-process provider and optional dependency would fit; done means local embedding libraries can be selected without requiring an HTTP client while keeping the core package lightweight.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- python
- Domain
- backend-api-design, machine-learning
- Issue type
- Feature
- Difficulty
- 5/5
- Estimated time
- Over a week
- Activity status
- Stale
- Clarity
- Mostly clear
- Newbie friendliness
- 25/100