It will be interesting to see where we are a couple of months from now, the pressure to release new models to catch up is rising. Which is a great thing, no one should be at the top for very long. Accelerate.
If I remember correctly, Dr. W-G, you claimed in the past few days that the “perfect” LLM is theorized to contain 10 quadrillion parameters. Am I dreaming this? And if not, where did this figure emerge, how soon will it be before it is reached, and finally, do you anticipate that it will indeed be “perfect”?
I would love to hear your opinion on this question. When will the race for AI development reach a conclusion? When will enough compute be enough? AGI? ASI? Or will it continue on forever? Will the shortage of power to run the data centers hold back this process? Maybe this is a good thing!
It will be interesting to see where we are a couple of months from now, the pressure to release new models to catch up is rising. Which is a great thing, no one should be at the top for very long. Accelerate.
Don’t Decel!
Alex- you keep crushing it with every update! Your emails are the highlight of my week and always a treasure to read.
Thanks!
If I remember correctly, Dr. W-G, you claimed in the past few days that the “perfect” LLM is theorized to contain 10 quadrillion parameters. Am I dreaming this? And if not, where did this figure emerge, how soon will it be before it is reached, and finally, do you anticipate that it will indeed be “perfect”?
I thought Alex suggested compression is intelligence, implying that the universal algorithm will be surprisingly simple (and low parameter).
I would love to hear your opinion on this question. When will the race for AI development reach a conclusion? When will enough compute be enough? AGI? ASI? Or will it continue on forever? Will the shortage of power to run the data centers hold back this process? Maybe this is a good thing!
Any info on how fast the “protected” models are moving ? Feel like the release models will be great but nowhere near the top end