As far as I understand, the training data is closed source. But, the methodology of training is open source which allows independent parties to recreate the model from scratch and see similar results. Not only can you download the full >400GB model using huggingface or ollama, but they also offer distilled versions of the model which are small enough to run on something like a raspberry pi. i’m running it locally on my machine at home with perplexica (perplexity.ai lookalike with searching capabilities)

If there is going to be an apk release, will there be support for Android TV as well? Thanks for the hard work!

Although I realize something like this might not be possible, i’d love (in a theoretical perfect world) a delegative/liquid federation. where you can “delegate” your blocklist be an aggregate of other people’s blocklist, which would allow a community of users independent of any admin to create a decentralized blocklist based upon mutual trust. To word it with an example, if I trust user A, who in turn trusts user B and C’s idea of who(/what communities) to block, i’ll then be blocking the same people as user B and C.

It could work in reverse too, if I trust user A who allows anime communities and user B who allows game communities, then I can see anime and game communities. If people trust me, they can see the same thing i’m seeing. Imo that would spur user interaction and make a decentralized way to not put any one person in power. If user B suddenly decides to only trust fascists, I don’t have to trust them anymore and those changes would be propagated.

I don’t know if that made sense, so sorry if that explanation is wack! It is loosely based on this concept that I read from awhile ago, for which I haven’t thought of the possible downsides.