DrYak on Nostr: npub1grx4w…suxfe what remains yo be seen is how much ressources go into using the ...
npub1grx4wzqldf8clxusxpz78adjcukkqvdkhf8amsdhr6sy3ctuu2fqvsuxfe (npub1grx…uxfe) what remains yo be seen is how much ressources go into using the model. (not training it, merely running it).
I wouldn't be surprised that it falls in the same "beyond diminishing returns" range as GPT-4, and that Sora requires a quite beefy datacenter to run; thus limiting how many people will be able to access it during the first couple of years.
I wouldn't be surprised that it falls in the same "beyond diminishing returns" range as GPT-4, and that Sora requires a quite beefy datacenter to run; thus limiting how many people will be able to access it during the first couple of years.