Skip to main content

​
Overview

Khoj uses an embedding model to understand documents. Multilingual embedding models improve the search quality for documents not in English. This affects both search and chat with docs experiences across Khoj.

​
Setup

To improve search and chat quality for non-English documents, you can use a multilingual model.
For example, the paraphrase-multilingual-MiniLM-L12-v2 supports 50+ languages, has decent search quality and speed for a consumer machine.
1

Configure Search Model

Open the search config on your server’s admin settings page. Either create a new search model, if none exists, or update the existing one.For example:
  • Set the bi_encoder field to sentence-transformers/paraphrase-multilingual-MiniLM-L12-v2
  • Set the cross_encoder field to mixedbread-ai/mxbai-rerank-xsmall-v1
2

Regenerate Content Index

This step is very important, as you’ll need to re-encode all your content with the new model.
Regenerate your content index from all the relevant clients.

​
Advanced Configuration

​
Query Prefix for Modern Models

Modern search/embedding models like mixedbread-ai/mxbai-embed-large-v1 expect a prefix to the query (or docs) string to improve encoding.
Update the bi_encoder_query_encode_config field of your embedding model with {prompt: <prefix-prompt>} to improve the search quality of these models. 
{
  "prompt": "Represent this query for searching documents"
}
You can pass any valid JSON object that the SentenceTransformer encode function accepts.
Languages: 50+ including Arabic, Chinese, Dutch, English, French, German, Italian, Korean, Polish, Portuguese, Russian, Spanish, TurkishPros:
  • Good balance of speed and quality
  • Works well on consumer hardware
  • Wide language support
Model: sentence-transformers/paraphrase-multilingual-MiniLM-L12-v2
Languages: Multilingual support with state-of-the-art performancePros:
  • Excellent search quality
  • Modern architecture
  • Supports query prefixes
Note: Requires more compute resourcesModel: mixedbread-ai/mxbai-embed-large-v1
Purpose: Rerank search results for improved relevancePros:
  • Small and fast
  • Improves final result quality
  • Works across languages
Model: mixedbread-ai/mxbai-rerank-xsmall-v1

​
Testing Your Configuration

After setting up multilingual support:
  1. Index some documents in your target language
  2. Try searching for content in that language
  3. Start a chat and ask questions about your documents
  4. Verify results are relevant and accurate
If search quality is not satisfactory, try a different embedding model or adjust the query prefix configuration for modern models.

Build docs developers (and LLMs) love

Get started for freeTalk to us