Perplexity CEO explains RAG: Retrieval-Augmented Generation | Aravind Srinivas and Lex Fridman

[@LexClips]

Full podcast episode: kzbin.info/www/bejne/m17KqKmjnd6IbaM Lex Fridman podcast channel: kzbin.info Guest bio: Arvind Srinivas is CEO of Perplexity, a company that aims to revolutionize how we humans find answers to questions on the Internet.

[@mackiej]

Great video. I wonder why there is no BM25Vectorizer in scikit-learn? There's a TfidfVectorizer.

[@SeenAndCheese]

haha I hope Lex video edited out what Aravind said about what not to say. That would be so metal. Meta-metal!

[@duongkstn]

From existential crisis channel to LLM teaching channel :)😅

[@manonamission2000]

when LLMs begin having 1,000,000,000 token windows, the RAG pattern might be obsolete

[@paultparker]

Indeed! I I Believe Perplexity is already doing this in normal, searches. The difficulty for very large context, windows is that the processing power required is the square of the context window size. my impression was that this is fundamental To how attention works in the transformer architecture, but it seems to me like there might be ways to optimize it with research.

[@djpuplex]

This is basically what AI is and will be for a while.

[@thegrandbizzare]

Obviously this was recorded before they were caught scraping Wired and Forbes articles, with robots.txt files, using unlisted ip addresses...lol

[@ChannelKit]

Perplexity is complete shit