Suggest updating rerank_model default in documentation from rerank-lite-1 to rerank-2.5

[ken1310-lang]

Documentation Improvement Request

Location:
On the “Complete Configuration Schema → Full JSON Configuration” section of the Configuration docs:
https://chunkhound.github.io/configuration/ (see snippet where "rerank_model": "rerank-lite-1")

Current configuration snippet:

"embedding": {
  "provider": "voyageai",
  "model": "voyage-3.5",
  "api_key": "pa-your-key",
  "base_url": "https://api.voyageai.com/v1",
  "rerank_model": "rerank-lite-1",
  "rerank_url": "/rerank"
}

Suggested change:

"embedding": {
  "provider": "voyageai",
  "model": "voyage-3.5",
  "api_key": "pa-your-key",
  "base_url": "https://api.voyageai.com/v1",
  "rerank_model": "rerank-2.5",
  "rerank_url": "/rerank"
}

Reason:
In my experimental usage, the rerank-lite-1 model yielded noticeably poorer semantic search / reranking results compared to the rerank-2.5 model (which is a newer/more capable model from the same provider). As this doc snippet is often used as a “default” for new users, updating it to the higher-accuracy model could give users a better out-of-the-box experience and reduce the need for manual overrides.

Impact:

• Improves default search/rerank accuracy for new users of ChunkHound.
• Reduces confusion: users may try the default and get sub-optimal performance, then assume other parts of the setup are broken.
• Keeps documentation aligned with best recommended practices for the embedding/reranking provider (in this case VoyageAI).

Thank you for the great project 🙏