3 ms·

What is "source" regarding an LLM? Public training data and initial parameters?

J_cst 3y ago
Not an expert, but often weights are mentioned as not being open sourced. Happy to get corrected, as I'm not really sure.
- blackeyeblitzar 3y ago
  Weights aren’t source because the goal of having open source software is that you can know how the software you’re consuming works, and you can produce the final software (the executable) using the source yourself. When you only have weights, you are getting something like the executable. Sure you can tweak it, but you don’t have the things you need to reproduce it or to examine how it works and validate it for your purposes. As such open weights are not in the spirit of open source.
  - robertlagrant 3y ago
    I don't think the previous commenter was saying that it's okay to only release the weights.
observationist 3y ago
The parameters and the license. Mistral uses Apache 2.0, a neatly permissive open source license. As such, it's an open source model. Models are similar to code you might run on a compiled vm or native operating system. Llama.cpp is to a model as Python is to a python script. The license lays out the rights and responsibilities of the users of the software, or the model, in this case. The training data, process, pipeline to build the model in the first place is a distinct and separate thing from the models themselves. It'd be nice if those were open, too, but when dealing with just the model: If it uses an OSI recognized open source license, it is an open source model. If it doesn't use an OSI recognized open source license, it's not. Llama is not open source. It's corporate freeware.
- littlestymaar 3y ago
  Mistral is not “open source” either since we cannot reproduce it (the training data is not published). Both are open weight models, and they are both released under a license whose legal basis is unclear: it's not actually clear if they own any intellectual property over the model at all. Of course they claim such IP, but no court has ruled on this yet AFAIK and legislators could also enact laws that make these public domain altogether.
  - jiggawatts 3y ago
    Is “reproducibility” actually the right term here? It’s a bit like arguing that Linux is not open source because you don’t have every email Linus and the maintainers ever received. Or that you don’t know what lectures Linus attended or what books he’s read. The weights “are the thing” in the same sense that the “code is the thing”. You can modify open code and recompile it. You can similarly modify weights with fine tuning or even architectural changes. You don’t need to go “back to the beginning” in the same sense that Linux would continue to be open source even without the Git history and the LKM mailing list.
    - littlestymaar 3y ago
      > It’s a bit like arguing that Linux is not open source because you don’t have every email Linus and the maintainers ever received. Or that you don’t know what lectures Linus attended or what books he’s read. Linux is open source, because you can actually compile it yourself! You don't need Linus's email for that (and if you needed some secret cryptographic key on Linus' laptop to decrypt and compile the kernel, then it wouldn't make sense to call it open-source either). A language model isn't a piece of code, it's a huge binary blob that's being executed by a small piece of code that contains little of the added value, everything that matters is in the blob. Sharing only the compiled blob and the code to run makes it unsuitable for an “open source qualifier” (It's kind of the same thing as proprietary Java code: the VM is open-source but the bytecode you run on it isn't). And yes, you can fine-tune and change things in the model weights themselves the same way you can edit the binary of a proprietary game to disable DRMs, that doesn't make it open-source either. Fine tuning doesn't give you the same level of control over the behavior of the model as the initial training does, like binary hacking doesn't give you the same control as having the source code to edit and rebuild.
blackeyeblitzar 3y ago
See this discussion and blog post about a model called OLMo from AI2 (https://news.ycombinator.com/item?id=39974374 https://news.ycombinator.com/item?id=39974374). They try to be more truly open, although here are nuances even with them that make it not fully open. Just like with open source software, an open source model should provide everything you need to reproduce the final output, and with transparency. That means you need the training source code, the data sets, the evaluation suites, the inference code, and more. Most of these other models, like Llama, are open weight not open source - and open weight is just openwashing, since you’re just getting the final output like a compiled executable. But even with OLMo (and others like Databrick’s DBRX) there are issues with proprietary licenses being used for some things, which prevent truly free use. For some reason in the AI world there is heavy resistance to using OSI-approved licenses like Apache or MIT. Finally, there is still a lack of openness and transparency on the training data sets even with models that release those data sets. This is because they do a lot of filtering to produce those data sets that happen without any transparency. For example AI2’s OLMo uses a dataset that has been filtered to remove “toxic” content or “hateful” content, with input from “ethics experts” - and this is of course a key input into the overall model that can heavily bias its performance, accuracy, and neutrality. Unfortunately, there is a lot missing from the current AI landscape as far as openness.
- sunandcoffee 3y ago
  what are you thoughts on projects like these: https://www.llm360.ai/ https://www.llm360.ai/ seems like they make everything available.
PeterisP 3y ago
I think the proper way is to start the consideration from the perspective of what the user can do, for example, from the four freedoms of FSF (https://www.gnu.org/philosophy/free-sw.html.en#four-freedoms https://www.gnu.org/philosophy/free-sw.html.en#four-freedoms) The starting point is the ability to run the LLM as you wish, for any purpose - so if a license prohibits some uses and you have to start any usage with thinking whether it's permitted or not, that's a fail. Then the freedom where "source" matters is the practical freedom to change the behavior so it does your computing as you wish. And that's a bit tricky - since one interpretation would require having the training data, training code and parameters; but for current LLMs the training hardware and cost of running it is a major practical limitation, so much that one could argue that the ability to change the behavior (which is the core freedom that we'd like) is separate from the ability to recreate the model, and would be more relevant in the context of the "instruction training" which happens after the main training, is the main determiner of behavior (as opposed to capability), and so the main "source would be the data for that (instruct training data, and the model weights before that finetuning) so that you can fine-tune the model on different instructions, which requires much less resources than training it from scratch, and don't have to start with the instructions and values imposed on the LLM by someone else.

Not an expert, but often weights are mentioned as not being open sourced. Happy to get corrected, as I'm not really sure.

blackeyeblitzar 3y ago
Weights aren’t source because the goal of having open source software is that you can know how the software you’re consuming works, and you can produce the final software (the executable) using the source yourself. When you only have weights, you are getting something like the executable. Sure you can tweak it, but you don’t have the things you need to reproduce it or to examine how it works and validate it for your purposes. As such open weights are not in the spirit of open source.
- robertlagrant 3y ago
  I don't think the previous commenter was saying that it's okay to only release the weights.

observationist 3y ago

The parameters and the license. Mistral uses Apache 2.0, a neatly permissive open source license. As such, it's an open source model. Models are similar to code you might run on a compiled vm or native operating system. Llama.cpp is to a model as Python is to a python script. The license lays out the rights and responsibilities of the users of the software, or the model, in this case. The training data, process, pipeline to build the model in the first place is a distinct and separate thing from the models themselves. It'd be nice if those were open, too, but when dealing with just the model: If it uses an OSI recognized open source license, it is an open source model. If it doesn't use an OSI recognized open source license, it's not. Llama is not open source. It's corporate freeware.

littlestymaar 3y ago
Mistral is not “open source” either since we cannot reproduce it (the training data is not published). Both are open weight models, and they are both released under a license whose legal basis is unclear: it's not actually clear if they own any intellectual property over the model at all. Of course they claim such IP, but no court has ruled on this yet AFAIK and legislators could also enact laws that make these public domain altogether.
- jiggawatts 3y ago
  Is “reproducibility” actually the right term here? It’s a bit like arguing that Linux is not open source because you don’t have every email Linus and the maintainers ever received. Or that you don’t know what lectures Linus attended or what books he’s read. The weights “are the thing” in the same sense that the “code is the thing”. You can modify open code and recompile it. You can similarly modify weights with fine tuning or even architectural changes. You don’t need to go “back to the beginning” in the same sense that Linux would continue to be open source even without the Git history and the LKM mailing list.
  - littlestymaar 3y ago
    > It’s a bit like arguing that Linux is not open source because you don’t have every email Linus and the maintainers ever received. Or that you don’t know what lectures Linus attended or what books he’s read. Linux is open source, because you can actually compile it yourself! You don't need Linus's email for that (and if you needed some secret cryptographic key on Linus' laptop to decrypt and compile the kernel, then it wouldn't make sense to call it open-source either). A language model isn't a piece of code, it's a huge binary blob that's being executed by a small piece of code that contains little of the added value, everything that matters is in the blob. Sharing only the compiled blob and the code to run makes it unsuitable for an “open source qualifier” (It's kind of the same thing as proprietary Java code: the VM is open-source but the bytecode you run on it isn't). And yes, you can fine-tune and change things in the model weights themselves the same way you can edit the binary of a proprietary game to disable DRMs, that doesn't make it open-source either. Fine tuning doesn't give you the same level of control over the behavior of the model as the initial training does, like binary hacking doesn't give you the same control as having the source code to edit and rebuild.

blackeyeblitzar 3y ago

See this discussion and blog post about a model called OLMo from AI2 (https://news.ycombinator.com/item?id=39974374 https://news.ycombinator.com/item?id=39974374). They try to be more truly open, although here are nuances even with them that make it not fully open. Just like with open source software, an open source model should provide everything you need to reproduce the final output, and with transparency. That means you need the training source code, the data sets, the evaluation suites, the inference code, and more. Most of these other models, like Llama, are open weight not open source - and open weight is just openwashing, since you’re just getting the final output like a compiled executable. But even with OLMo (and others like Databrick’s DBRX) there are issues with proprietary licenses being used for some things, which prevent truly free use. For some reason in the AI world there is heavy resistance to using OSI-approved licenses like Apache or MIT. Finally, there is still a lack of openness and transparency on the training data sets even with models that release those data sets. This is because they do a lot of filtering to produce those data sets that happen without any transparency. For example AI2’s OLMo uses a dataset that has been filtered to remove “toxic” content or “hateful” content, with input from “ethics experts” - and this is of course a key input into the overall model that can heavily bias its performance, accuracy, and neutrality. Unfortunately, there is a lot missing from the current AI landscape as far as openness.

sunandcoffee 3y ago
what are you thoughts on projects like these: https://www.llm360.ai/ https://www.llm360.ai/ seems like they make everything available.

PeterisP 3y ago

I think the proper way is to start the consideration from the perspective of what the user can do, for example, from the four freedoms of FSF (https://www.gnu.org/philosophy/free-sw.html.en#four-freedoms https://www.gnu.org/philosophy/free-sw.html.en#four-freedoms) The starting point is the ability to run the LLM as you wish, for any purpose - so if a license prohibits some uses and you have to start any usage with thinking whether it's permitted or not, that's a fail. Then the freedom where "source" matters is the practical freedom to change the behavior so it does your computing as you wish. And that's a bit tricky - since one interpretation would require having the training data, training code and parameters; but for current LLMs the training hardware and cost of running it is a major practical limitation, so much that one could argue that the ability to change the behavior (which is the core freedom that we'd like) is separate from the ability to recreate the model, and would be more relevant in the context of the "instruction training" which happens after the main training, is the main determiner of behavior (as opposed to capability), and so the main "source would be the data for that (instruct training data, and the model weights before that finetuning) so that you can fine-tune the model on different instructions, which requires much less resources than training it from scratch, and don't have to start with the instructions and values imposed on the LLM by someone else.