Ricerche recenti
Opzioni di ricerca
Amministrato da:
mastodon uno admin 
(A piece of advice: better start mirroring those #DeepSeek GitHub repositories with code and model data before they "accidentally" disappear. I am quite sure someone is already filing DMCA requests and preparing lawsuits for whatever weird copyright infringement they can come up with. Remember the old song: It's only wrong when THEY do it ;)
@jwildeboer
So we shall mirror it #anonymously and over a #VPN plus tunneling into #tor then #i2p then IP over Avian Carriers… 
Jokes apart how big is it?
And even more seriously, isn't it "a little #biased" at least on the #political issues?
The base tensors are nearly 700 GB if I'm right.
@Ollivdb
Quite a lot of (#biased) data, I dare say… I'd rather get the code. That's where the real innovation lies, if they really found a way to train a "neural net" with a thousand fold efficiency compared to the incumbents
@jwildeboer