> But I feel the only reason Chinese models have been able to gain such popularity is BECAUSE they are open source.
Why do people keep calling these "open source" models? They're not that different from NVIDIA's binary blobs. Sure, they work, and you can run them and get results, but the source code, training data, and training process are largely hidden and proprietary. Publishing weights is not the same as publishing the source. They're open-weight models, not open-source models.
I'd argue that they're still open-source because inference code is fully public. In operation, a user can crack open the model and examine its internal state, and we also see that many people produce various modifications or fine-tunings of public models for their own goals.
In my mind, this meets the conceptual and technical bars of an 'open source' designation. We can also imagine the opposite, where inference itself has to pass via an inscrutable binary blob that keeps the model's technology stack fully proprietary.
You're correct that we're missing the training data, the training recipe/curriculum, and any secret sauce that went into optimization of same, but I don't think that's an essential feature of an open source designation.
Instead, I think that the model weights are a kind of rendered artifact, and it should be evaluated on that basis. We don't treat permissively licensed images as something lesser, even if the photographer or illustrator supplies only the final image rather than the full camera raw file and/or Photoshop layer stack.
The difference, of course, is that the rendered artifact (weight blob) in a language model is so much more important than a typical photograph.
Nah this is like saying that Photoshop is open source just because you can download and run the binary on your own computer instead of through a SaaS webapp. I think basically everyone would disagree with that.
> In operation, a user can crack open the model and examine its internal state, and we also see that many people produce various modifications or fine-tunings of public models for their own goals.
I mean all of open source software (as we traditionally defined it) relies on an ISA, which is actually executed by microcode that isn’t available to the public?
I don’t see how this is any different. It’s like saying Intel has to publish their transistor physical library and microcode to be “truly” open source. Like I don’t have the internal notes about the most recent microcode changes (which is akin to “training process” in the analogy).
Why do people keep calling these "open source" models? They're not that different from NVIDIA's binary blobs. Sure, they work, and you can run them and get results, but the source code, training data, and training process are largely hidden and proprietary. Publishing weights is not the same as publishing the source. They're open-weight models, not open-source models.