Just shipped RagPack v1.0 π
RagPack is an open source lightweight RAG infrastructure for developers that want a cost efficient solution.
Designed with Ollama first experience, works with OpenAI as well.
https://t.co/w2Gcelzj6s
#buildinpublic#opensource#rag#ai#ollama
@the_real_ori@kinmorychin In memory embeddings is a different decision here, you would need to re-embed all the chunks after each build or restarts. How many documents do you typically have and how big are they?
@SidDegen@shedntcare_ Hmm no docker part is interesting definitely. Surely it would cause issues when you try to deploy to different environments no? I wonder how are they solving that problem.
You can always self-host lightweight RAG infras btw.
@rihntv It's tough for sure. The expectations have risen as much as the tools have progressed. But you can still use them effectively to make a difference.
Heads up if you're on Vercel's free tier. Recently got an email about reaching my free usage quota, one route was burning ~3 min of Active CPU per billing window.
Fix: server-side caching with Next.js's unstable_cache. More below..
https://t.co/5QZldxH7uX