Allow use of Custom Embedding and Rerank models

Hi,
Thank you for making this project, I've been enjoying using it and it works well. 

I'd like to be able to use custom models, I've been eying the nvidia nemotron embeder and reranker (nvidia/llama-embed-nemotron-8b).  What was the methodology for picking the current models?