Nvidia puts 30 years of high value knowhow in a 13B LLM

They seem to be using RAGs to prevent hallucinations (which I imagine would be really bad in their context).

The 13B Llama 2 model is not that good compared to the best ones. Maybe it was easier for them to fine tune?

This reminds me of a Star Trek (The Next Generation) character consulting an AI expert on the Holodeck.

how much you think china is investing in industrial espionage for this.