MLOps question - Setting up enviornment for RAG application

Greetings fellow Coders

I’m familiar with RAG applications conceptually.

I decided to start an experiment and build one myself.

I have all the code; but I am having trouble building the environment. It keeps crashing. I’m using VSCode locally on a MAC M1max, fairly new with plenty of horse power.

I have a hypothesis, where I am creating environments and my packages have conflicts. I will either have to troubleshoot them one by one, or…

I am leaning towards running the code in a google colab enviornment or setup a sandbox for experiments in GCP.

Does anyone reflect on their own experience setting up your first RAG?

Hi Dan. I built a simple RAG when I was learning. It’s a system to search for research papers on Arxiv using natural language. It uses FAISS, Sentence Transformers and OpenAi. It might help you understand how to put all the building blocks together and get up and running. I’ve also included a list of learning resources that I found helpful.

I used sentence-transformers with documents stored in a chromeDB database. I tried using a different dataset, using polars data frame and the kernel kept crashing. I created 2 new environments, but they keep crashing too. I can troubleshoot, but it seems very time consuming vs. using a cloud solution like colab.