3 comments
A posttraining data commons and the infrastructure surrounding it is the next great open source project IMO, akin to Postgres or Linux.<p>The Marin project from Stanford (and I believe now ran by Openathena, the authors of this blog post) is really promising and cool: <a href="https://marin.community/" rel="nofollow">https://marin.community/</a>
I think a post-training common would have to be somewhat protected, since these things would be created with great effort and would potentially impact people's ability to earn money in their jobs.<p>I could also imagining them happening as state projects for the sake of competitiveness.
I'm not sure I'll do it justice, but one of the general ideas behind open-model initiatives is that the economic impacts of specialized post-training are a fait accompli. OpenAI and Anthropic have spent $many-millions procuring closed posttraining data.<p>The relevant question here is whether you want open-weight models — projects like Open Athena — to remain or become competitive in the same range of specialties.
They're already kinda protected, open weight releases rarely come with datasets, especially post-training datasets. NVIDIA is probably the exception.<p>I also wonder when the NSA starts sharing their wiretaps/etc with labs for 'national security'. Probably already happening, to be honest.
> I also wonder when the NSA starts sharing their wiretaps/etc with labs for 'national security'.<p>What would this benefit the NSA? If they share wiretap records with labs such as Anthropic, then suddenly anyone who can jailbreak a model gets access to this data. You don't need to train a model on data to understand or interpret it, as far as I know - providing others (esp. AI labs) with sensitive data is a breach of security, or I am not understanding.
[dead]