1,360 karma · joined April 24, 2016
It must be like a pit fight, but with connections and half truths
He talked about the people getting their homework and bachelor's done.
Pretty much crushed my worldview. Cheating is, was and will be rampant.
Because they auto route still your requests, you never know if something costs one coin or a tenth of a coin
But I think the next step will be even more thinking on smaller models. Maybe fine-tuned and we get really crazy stuff
I don't get this thread.... Really. Is it full of bots?
That's at least my impression. Models didn't get get better, just more thinking and testing and sometimes fixing things you didn't ask for ( hello opus, can you check xxx, opus: I fixed it..)
Next step is a model with 10 GB thinking for ten minutes.
No one will overpay to be nice
Or on runpod with rented gpus or inference.
And it was damn cheap too! The main page cost was like 5 dollar.
Models are too big now
Those implementations don't have access to the low level encoding Nvidia has.
There is a rumour, that they did this because of GeForce now, because there never will an open source implementation/ service competition for GeForce now.
Absolute insanity
You either do it, or you are out
Still cheaper then being jobless or without the bachelor
Healthy in a way, but surprisingly a lot of sugar ( either added or from the fruits)
Missed half the stuff the other half was outdated/ didn't verify.
How?
Which model is behind it?