RAG setup : all-MiniLM-L6-v2

[delta-whiplash]

Hello,

First, I would like to extend my gratitude to everyone who has contributed to this project and helped me improve my workflow daily. I am pleased to announce that I have successfully implemented the RAG setup on my system, and it is currently functioning well.

At the moment, I am utilizing the "all-MiniLM-L6-v2" models for embedding, as suggested in the documentation. I am considering increasing the precision by using other embedding models available in the repo, such as:

• https://huggingface.co/sentence-transformers/msmarco-MiniLM-L-6-v3
• https://huggingface.co/sentence-transformers/all-MiniLM-L12-v2

As a novice in embedding, I would appreciate it if someone could explain the impact and major differences between these models. Specifically, I am interested in understanding how these models differ in terms of:

1. Size: What are the file sizes of each model, and how do they affect the computational resources required in terms of memory and processing power?
2. Performance: How do the models compare in terms of accuracy, speed, and any trade-offs between them?
3. Applications: Under what scenarios would one model be more suitable than the other?

Your insights and expertise would be much appreciated.

Thank you <3