Local Code Search Embeddings (Ollama) FAQs

Question 1

How do I update my index after switching models?

Accepted Answer

If you change your embedding model or dimensions, you must delete your existing index file (typically .grepai/index.gob) and run the 'grepai watch' command to re-index your project.

Question 2

Which embedding model is best for general code search?

Accepted Answer

The nomic-embed-text model is highly recommended as it provides the best balance of speed, low memory footprint (approx. 274MB), and excellent quality for codebases.

Question 3

Can I run Ollama on a different machine than my code?

Accepted Answer

Yes, you can configure a remote Ollama server by setting the endpoint in your .grepai/config.yaml to the server's IP address and ensuring the OLLAMA_HOST is set to 0.0.0.0 on the host.

Question 4

Why should I use Ollama for GrepAI embeddings?

Accepted Answer

Ollama ensures 100% privacy by keeping your code on your local machine, eliminates recurring API costs, and allows for offline semantic search with zero network latency.

Question 5

Does this skill support GPU acceleration?

Accepted Answer

Yes, Ollama automatically leverages Metal on Apple Silicon and CUDA on NVIDIA GPUs to significantly speed up the embedding generation process.

Local Code Search Embeddings (Ollama)

Key Features

Use Cases

Local Code Search Embeddings (Ollama)

Key Features

Use Cases