Artificial Intelligence thread
- Thread starter 9dashline
- Start date
Some o3 commentary
Basically it seems the test creator is an openAI shill and intentionally omitted a few things that put the credibility of the test in jeopardy.
Hearing quite a few US analysts + engineers confused and coping this morning.
Basically it seems the test creator is an openAI shill and intentionally omitted a few things that put the credibility of the test in jeopardy.
This is insanity. It is trained with 10% the compute it took to train LLama 405.
Hearing quite a few US analysts + engineers confused and coping this morning.
Last edited:
I had a SORA feeling about o3..... it was supposed to be part of "shipmas" but not only was o3 not shipped, it was announced with only benchmarks and not even a live demonstration.... the only reason Altman would do that is because he had something to hide and was in fake it until make it mode, otherwise he would be bragging with real live examples... instead of just a rigged chart/ fake benchmark etc... and they even told the ARC AGI dude to not disclose how much compute was used and the hardware/cost lol.... probably millions of dollars for a fake horse and pony show
And after I purchased the o1 pro mode for $200, I really expected, wanted and hoped that pro version of SORA would be way better than Kling, but in all my tests, its worse than Kling...
Last edited:
wait, this is that annihilation dude? lmao
CCP does not micromanage the nation at this level.
Are you using the free or paid version of Kling?
Yes, I saw that, but still quite unbelievable to do this.
My point is that it's in Deepseek's interest and China's interest to not fully acknowledge just how many GPUs they have or are using to training. So, I do put a Question mark in stuff like this.