This discussion is so dumb - finetuning a base model costs ~$1 with LORA/QLORA and can yield same performance as gpt-4, but at 1/100 of the cost per token.
I had opened a new tab back when this comment was just a few minutes old in hopes that when I came back there was some really great blog post linked with the details on the sorcery.
lol Emad was always seemed like an obvious fraud to me. Not quite SBF level but same vibe. Whenever someone goes overboard on the nerd look it’s always a red flag.
Whoever is dumb enough to invest in a company that hasn’t even bothered to come up with their own company name, but just copied the name of the current most hyped competitor in their category, deserves to lose their money.
Not really disagreeing with your insight, but on the flip side of this, I bought the stock precisely because of the ticker name (when it changed from VTIQ to NKLA) and sold it shortly thereafter for a significant upside.
Depends on your definition of winning - special-purpose tuning is vastly more cost effective since it allows you to train a smaller model that can perform specific tasks as good as a bigger one.
A good analogy is building a webapp - would you prefer to hire a developer with 30+ years experience in various CS domains as well as a PhD or a specialized webdev with 5 years experience at a tenth of the rate?
