4 ms·
I do wonder if you actually need two models here. Audio-to-audio hindbrain on the client, and a beefy text-mode frontal lobe somewhere in the cloud, with the co
by regularfry 5mo ago
I do wonder if you actually need two models here. Audio-to-audio hindbrain on the client, and a beefy text-mode frontal lobe somewhere in the cloud, with the comms between them explicitly trained in as a potentially low-bandwidth steering connection transferring embeddings, not text.
- 6510 5mo agohttps://en.wikipedia.org/wiki/Sloot_Digital_Coding_System https://en.wikipedia.org/wiki/Sloot_Digital_Coding_System