For those in #Singapore and interested in #datavisualization & #llms, I'm talking about Visualizing LLM Hallucinations at SUTD on Thu 8 Feb at 7 pm SGT.

This is for a non-technical audience. We'll visualize the basics of how LLMs work, how they make mistakes, and at least one technique on how to spot these.

https://lnkd.in/g-AYMaMi

Incredible, Rishabh Srivastava!

I'm curious: have you benchmarked with https://github.com/vanna-ai/vanna on SQL generation?

What's a good similarity threshold when comparing text?

When using #embeddings from text-embedding-ada-002 from #OpenAI, I used ~85% cosine similarity as to indicate good similarity.

With the new text-embedding-3 models, I'm re-calibrating my mental model. Anything with a cosine similarity of ~45% indicates good similarity.

https://lnkd.in/gizza77A