Highly misleading. They finetuned an existing model using a different existing model in a process called distillation.
The article is effectively saying “our model only cost $50 to make, plus whatever tens or hundreds of millions of dollars the models we stole from cost.”
while absolutely true, the same can be said about my chinese nonsensename companys dehumidifier that I bought for 1/4 of the cost of an american brand name one
It’s pretty much how the global economy has worked for a few decades, right? Advanced countries design, research and run things, developing countries build them.
I’m not really sure what it has to do with OP, though.
That would only be a valid comparison if the american brand dehumidifier, as a complete product, was a part of the chinese one’s bill of materials. This is closer to the cartoon meme image where McGuyver builds a megaphone out of a squirrel, twigs, and a megaphone.
That’s rookie numbers I trained one in 1min with $1!
How long before congress bans this one too
The underlying research story is interesting, but the way it’s written up actively makes it worse.
The researchers based s1 on Qwen2.5, an open-source model from Alibaba Cloud.
Watch me create a racing car for less than $50. Step 1: start with a Mercedes F1 racer…
Trained it to do very basic arithmetic tasks, not to rival OpenAI.
