embeddings

A Module providing a unified interface for generating text embeddings — with first-class attention to Arabic text — across five backends: local sentence-transformers models, a dedicated Arabic-tuned local pipeline, and three hosted APIs (OpenAI, Google Gemini, Mistral AI), plus a fully local Ollama server backend.

[!NOTE] About this documentation This documentation was generated directly from static analysis of the embeddings package source (embeddings/*.py). Every signature, parameter, default value, and behavior described here is taken from the actual implementation. Where the implementation is ambiguous, incomplete, or appears to contain a defect, that is called out explicitly rather than guessed at — see Known Issues.

The package identifies itself only as embeddings in code (see embeddings/__init__.py); no "zerda" branding, CLI, REST API, plugin system, ORM layer, or test suite exists anywhere in the analyzed source, so those sections are not present in this documentation.

What's in the package

The package exposes eight public symbols from embeddings/__init__.py:

All five concrete embedders subclass BaseEmbedder and implement the same two required members: encode() and the embedding_dim property. This means any embedder can generally be swapped for another without changing calling code, modulo constructor arguments (API keys, etc.).

Where to go next

• New to the package? Start with Installation and Quickstart.
• Want the big picture? See Architecture.
• Configuring defaults? See Configuration.
• Looking for a specific class or method? See the API Reference.
• Hit something confusing in the source? Check FAQ & Known Issues.