Will need hands on to truly tell, but the doom demo seems very promising. If it can play that with text descriptions of where stuff is by distance and degrees in a 3D context then many GUI automation tasks should be easily doable
Ozzie_osman 10 hours ago||
This is quite the paradigm shift. Can't wait to get my hands on it.
mmastrac 15 hours ago||
Is this a Markov/Diffusion model with some sort of external Engram memory? If so, this could be extremely interesting.
gok 20 hours ago||
So... a classifier model?
dozerly 15 hours ago||
Very cool. LLMs have been borderline unusable as functions for the longest time, very excited for this direction.
altcognito 17 hours ago||
If it is so cheap, why such a limited release?
entrep 21 hours ago||
This puts the human even more out of the loop I'll guess?
zenlikethat 19 hours ago|
That's kinda the goal. Imagine all the automation in the world being able to embed intelligence directly inside it - factories could route based on more complicated questions, hardware could anticipate your needs. Customer support could be done without humans 90% of the time.
bthornbury 21 hours ago||
Is the tradeoff of the parallel output that we don't get arbitrary string generation? like output # of tokens is fixed ahead of time?
Either way, really cool and impressive.
zenlikethat 19 hours ago|
Yeah, it doesn't output strings, just decisions/answers.
copperx 18 hours ago||
Non-hallucinated ones at that.
bananaflag 21 hours ago|
Funny how it can do everything but not chat. Sort of how when I was a kid I thought of a medicine that could cure any disease except the common cold.