Open source is the core of 90% of the world's software, an unequivocal force for good, and yet it's under attack. Big companies and activists want to make open source AI illegal. They must be stopped.
Very impassioned plea here! I’m curious to understand how you view a couple elements though (that never seem to get discussed much)
1) An ML model is an opaque set of weights derived from training code and training data. But yet very few of the open source models release the training code/data. An ML model feels like a free binary-only compiler: You can use it for whatever you want, but you can’t make it yourself from scratch. I’m curious which aspects of open source you care about?
2) There are a lot of models being released with open weights not actually under open source. Llama et al come to mind. If you’re making an appeal to Linux and Red Hat, i’m sure you must be more familiar with the variants of open source licensing than I am. I know there are some open source advocates very upset at the dilution and cooptation of “open source”. So lI am curious if you arguing open weights, or actual true open source above?
https://openai.com/blog/openai-announces-leadership-transition
Very impassioned plea here! I’m curious to understand how you view a couple elements though (that never seem to get discussed much)
1) An ML model is an opaque set of weights derived from training code and training data. But yet very few of the open source models release the training code/data. An ML model feels like a free binary-only compiler: You can use it for whatever you want, but you can’t make it yourself from scratch. I’m curious which aspects of open source you care about?
2) There are a lot of models being released with open weights not actually under open source. Llama et al come to mind. If you’re making an appeal to Linux and Red Hat, i’m sure you must be more familiar with the variants of open source licensing than I am. I know there are some open source advocates very upset at the dilution and cooptation of “open source”. So lI am curious if you arguing open weights, or actual true open source above?
> You have a 1 in 93 chance of dying in an accident every time you get in a car.
Umm... I assume this is some sort of error in wording? Most Americans get in a car way way more than a hundred times in their lives without dying.
This was really interesting. Thank you!