#9 Add the ability to use Ollama Embeddings models

[sighmon]

Resolves #9

So that we can work completely off-line and with open source models, let's add OllamaEmbeddings.

Acceptance Criteria

• Add the ability to use OllamaEmbeddings

Relevant design files

• None

Testing instructions

1. Delete your vector database: rm -rf works_db
2. Add below to your config.env
3. Start your ollama server: OLLAMA_HOST=0.0.0.0:11434 ollama serve
4. Build a local virtual environment: make build-local
5. Start the chat to re-build embeddings using llama3 from our public API: make up-local (note creating embeddings takes a while)
6. Quit the chat, and start the server: make server-local
7. Note you can chat off-line: http://localhost:8000/playground/
8. Note you can also set a different embedding model to the llm model: EMBEDDINGS_MODEL=nomic-embed-text
9. Install that embedding model: ollama pull nomic-embed-text
10. Delete your vector database again: rm -rf works_db
11. Re-build the embeddings: make up-local

MODEL=llama3
LLM_BASE_URL=http://localhost:11434

[sam-s-maher]

Nice one, looks good