Ghodsi: Smaller custom models can beat large LLMs on domain accuracy
Ali Ghodsi · AI Food Fights in the Enterprise with Databricks' Ali Ghodsi · Sep 25, 2023 · at 7:29
Ali Ghodsi, CEO and co-founder of Databricks, discusses why enterprises prefer domain-specific custom models over giant general foundation models for specific operational tasks.
0:00 / 0:20exact quote · 20.1s
▶ Watch the full episode on YouTube → 720p mp4 · rendered on demand · StarZero watermark
“And there you're better off if you have a good data set to train, you can train a smaller model. The latency will be faster to use it later, and it will be cheaper to use it later, and yes, you can have absolutely accuracy that beats the really large model, but that very model that you built can't also entertain you on the weekend, and answer physics question, and help your kids do their homework.”
quote is from the automated transcript, cleaned for reading: filler sounds and stutters are removed, nothing is rephrased. names can be misheard (the analysis reads context, assessments check outside sources). how →
More from Ali Ghodsi
Ghodsi: Product-led growth is overrated for enterprise software
“This PLG thing is just overrated for the kind of business we have in the kind of market we are in.”
Ghodsi: All current LLM evaluation benchmarks are completely bullshit
“I kind of think all the benchmarks are bullshit, and so all these, so all the LLM benchmarks, here's how it works.”
Horowitz warned Ghodsi that selling Databricks early would cause lifelong regret
“Hey, you can do whatever you want. You can sell. You're going to make a lot of money and you'll be super successful in life. But you know, if you're like me, you're going to look back the rest of your life thinking, you know, I missed that one shot. That was t…”
Ghodsi: A commercial open-source company's biggest enemy is its free project
“But the problem was that you know, as it is often with open sources that you just everyone's just downloading the open source version. Actually your biggest enemy is your open source project.”
Ghodsi: Founders fail by hiring their own archetype for unfamiliar roles
“They'll do well in their own archetype. When they have to step outside of their own archetype, they make a mistake. They hire people that are like their own archetype in other roles where that could be lethal.”
Ghodsi: Databricks passed on acquiring a successful company over weak talent
“There is one really good company that well, one really successful company that we never acquired. And I always vetoed it whenever it came up because I just think that the quality of their employee base is not great. And I didn't want it to dilute Databricks.”