It's not open source. The provide the model and the weights, but not the source code and, crucially, the training data. As long as LLM makers don't provide the training data (and they never will, because then they will be admitting to stealing), LLMs are never going to be open source.
(a) You have everything you need to be able to re-create something, and at any step of the process change it.
(b) You have broad permissions how to put the result to use.
The "open source" models from both Meta so far fail either both or one of these checks (Meta's fails both). We should resist the dilution of the term open source to the point where it means nothing useful.
Agreed, but the "connotations don't match" is mostly because the folks who chose to call it open source wanted the marketing benefits of doing so. Otherwise it'd match pretty well.
At the risk of being called rms, no, that's not what open source means. Open source just means you have access to the source code. Which you do. Code that is open source but restrictively licensed is still open source.
That's why terms like "libre" were born to describe certain kinds of software. And that's what you're describing.
This is a debate that started, like, twenty years ago or something when we started getting big code projects that were open source but encumbered by patents so that they couldn't be redistributed, but could still be read and modified for internal use.
> Open source just means you have access to the source code.
That's https://en.wikipedia.org/wiki/Source-available_software , not 'open source'. The latter was specifically coined [1] as a way to talk about "free software" (with its freedom connotations) without the price connotations:
The argument was as follows: those new to the term "free software" assume it is referring to the price. Oldtimers must then launch into an explanation, usually given as follows: "We mean free as in freedom, not free as in beer." At this point, a discussion on software has turned into one about the price of an alcoholic beverage. The problem was not that explaining the meaning is impossible—the problem was that the name for an important idea should not be so confusing to newcomers. A clearer term was needed. No political issues were raised regarding the free software term; the issue was its lack of clarity to those new to the concept.
It's common for terms to have a more specific meaning when combined with other terms. "Open source" has had a specific meaning now for decades, which goes beyond "you can see the source" to, among other things, "you're allowed to it without restriction".
I don't know why you've been downvoted. This is a 100% correct history. "Open source" was specifically coined as a synonym to "free software", and has always been used that way.
> Open source just means you have access to the source code. Which you do.
No, they also fail even that test. Neither Meta nor DeepSeek have released the source code of their training pipeline or anything like that. There's very little literal "source code" in any of these releases at all.
What you can get from them is the model weights, which for the purpose of this discussion, is very similar to compiler binary executable output you cannot easily reverse, which is what open source seeks to address. In the case of Meta, this comes with additional usage limitations on how you may put them to use.
As a sibling comment said, this is basically "freeware" (with asterisks) but has nothing to do with open source, either according to RMS or OSI.
> This is a debate that started, like, twenty years ago
For the record, I do appreciate the distinction. This isn't meant as an argument from authority at all, but I've been an active open source (and free software) developer for close to those 20 years, am on the board of one of the larger FOSS orgs, and most households have a few copies of FOSS code I've written running. It's also why I care! :-)
The weights, which are part of the source, are open. Now you are arguing it not being open source because they don't provide the source for that part of the source. If you follow that reasoning you can ad infinitum claim the absence of sources since every source originates from something.
The source is the training data and the code used to turn the training data _into_ the weights. Thus GP is correct, the weights are more akin to a binary from a traditional compiler.
To me this 'source' requirement does not make sense. It is not that you bring training data and the application together and press a train button, there's much more actions involved.
Also the training data is of a massive amount.
Additionally, what about human in the loop training, do you deliver humans as part of the source?
> they also fail even that test. Neither Meta nor DeepSeek have released the source code of the
This debate is over and makes the open source community look silly. Open model and weights is, practically speaking, open source for LLMs.
I have tremendous respect for FOSS and those who build and maintain it. But arguing for open training data means only toy models can practically exist. As a result, the practical definition will prevail. And if the only people putting forward a practical definition are Meta et al, this is what you get: source available.
I'm not arguing for open training data BTW, and the problem is exactly this sort of myopic focus on the concerns of the AI community and the benefits of open-washing marketing.
Completely, fully breaking the meaning of the term "open source" is causing collateral damage outside the AI topic, that's where it really hurts. The open source principle is still useful and necessary, and we need words to communicate about it and raise correct expectations and apply correct standards. As a dev you very likely don't want to live in a tech environment where we regress on this.
It's not "source available" either. There's no source. It's freeware.
"I can download it and run it" isn't open source.
I'm actually not too worried that people won't eventually re-discover the same needs that open source originally discovered, but it's pretty lame if we lose a whole bunch of time and effort to re-learn some lessons yet again.
> it's pretty lame if we lose a whole bunch of time and effort to re-learn some lessons yet again
We need to relearn because we need a different definition for LLMs. One that works in practice, not just at the peripheries.
Maybe we can have FOSS LLMs vs open-source ones, like we do with software licenses. The former refers to the hardcore definition. The latter the practical (and widely used) one.
Sure, I don't disagree. I fully understand the open-weights folks looking for a word to communicate their approach and its benefits, and I support them in doing so. It's just a shame they picked this one in - and that's giving folks a lot of benefit of the doubt - a snap judgement.
> Maybe we can have FOSS LLMs vs open-source ones, like we do with software licenses.
Why not just call them freeware LLMs, which would be much more accurate?
There's nothing "hardcore" or "zealot" about not calling these open source LLMs because there's just ... absolutely nothing there that you call open source in any way. We don't call any other freeware "open source" for being a free download with a limited use license.
This is just "we chose a word to communicate we are different from the other guys". In games, they chose to call it "free to play (f2p)" when addressing a similar issue (but it's also not a great fit since f2p games usually have a server dependency).
> Why not just call them freeware LLMs, which would be much more accurate?
Most of the public is unfamiliar with the term. And with some of the FOSS community arguing for open training data, it was easy to overrule them and take the term.
Most of the public is also unfamiliar with the term open source, and I'm not sure they did themselves any favors by picking one that invites far more questions and needs for explanation. In that sense, it may have accomplished little but its harmful effects.
I get your overall take is "this is just how things go in language", but you can escalate that non-caring perspective all the way to entropy and the heat death of the universe, and I guess I prefer being an element that creates some structure in things, however fleeting.
The only practical and widely used definition of open source is the one known as the Open Source Definition published by the OSI.
The set of free/libre licenses (as defined by the FSF) is almost identical to the set of open sources licenses (as defined by the OSI).
The debate within FOSS communities has been between copyleft licenses like the GPL, and permissive licenses like the MIT licence. Both copyleft and permissive licenses are considered free/libre by the FSF, and both of them are considered open source by the OSI.
People say this, but when it comes to AI models, the training data is not owned by these companies/groups, so it cannot be "open sourced" in any sense. And the training code is basically accessing that training data that cannot be open sourced, therefore it also cannot be shared. So the full open source model you wish to have can only provide subpar results.
They could easily list the data used though.
These datasets are mostly known and floating around.
When they are constructed, instructions for replication could be provided too
But I think my argument still stands though? Users can run Deepseek locally, so unless the US Gov't wants to reach for book burning levels or idiocy, there is not really a feasible way to ban the American public of running DeepSeek, no?
Yes, your argument still stands. But I think it's important to stand firm that the term "open source" is not a good label for what these "freeware" LLMs are.
