wishtrepreneur@alien.topBtoEntrepreneur@indiehackers.space•If you think all great ideas are taken already, read thisEnglish
1·
10 months agoNovelty is for amateurs, imo.
hey, don’t diss my fidget spinner! I made a whole $200 from dropshipping that!
why can’t you just train the “router” LLM on which downstream LLM to use and pass the activations to the downstream LLMs? Can’t you have “headless” (without encoding layer) downstream LLMs? So inference could use a (6.5B+6.5B) params model with the generalizability of a 70B model.