Doc: Update README examples to prevent double-wrapping of embedding functions

[danielaskdd]

Fix: Update README examples to prevent double-wrapping of embedding functions

Problem:
The EmbeddingFunc.__call__ method now contains processing logic that prevents functions from being wrapped twice. This causes issues when embedding functions already decorated with @wrap_embedding_func_with_attrs are wrapped again using EmbeddingFunc().

Root Cause:
Since openai_embed, ollama_embed, and similar functions are already decorated with @wrap_embedding_func_with_attrs, wrapping them again with EmbeddingFunc() results in double decoration, which breaks the expected behavior.

Solution:
Update all README examples that incorrectly use EmbeddingFunc() to wrap pre-decorated embedding functions. The correct approach is to either:

1. Use @wrap_embedding_func_with_attrs decorator to create a custom embedding function that calls xxx_embed.func (the unwrapped function)
2. Pass the decorated embedding function directly to LightRAG without additional wrapping

Before (Incorrect - Double Wrapping):

embedding_func=EmbeddingFunc(
    embedding_dim=768,
    func=lambda texts: ollama_embed(texts, ...)  # ollama_embed is already decorated!
)

After (Correct):

@wrap_embedding_func_with_attrs(embedding_dim=768, max_token_size=8192)
async def embedding_func(texts: list[str]) -> np.ndarray:
    return await ollama_embed.func(texts, ...)  # Call .func to access the unwrapped function

rag = LightRAG(
    ...
    embedding_func=embedding_func,  # Pass directly, no EmbeddingFunc wrapper needed
)

Key Points:

• Call xxx_embed.func instead of xxx_embed to avoid triggering double decoration
• The .func attribute accesses the underlying function while preserving the built-in retry logic
• Pass decorated functions directly to embedding_func parameter without additional EmbeddingFunc() wrapper

[danielaskdd]

@codex review

[chatgpt-codex-connector]

Codex Review: Didn't find any major issues. Can't wait for the next one!

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".