I think it’s nice.
ML4 was trained on 3,800 NVIDIA Grace Blackwell GPUs in our European datacenters -- our cluster in Bruyères-le-Châtel built using our Series B fundraise.
We are investing heavily in infrastructure, with our Series C and D clusters coming online soon. This additional compute will rsue more ambitious post-training, and iterate faster.
We expect large and rapid improvements in the weeks and months ahead.
2/n
1
0
2
250
@true_jae_yang > drive 1990s peugeot at 93mph down country lanes 3" wider than the car to office
> stop at boulangerie for baguette enroute
> eur2 bread is better than anything in a 5-acre wal-mart
> office has no a/c, is next to heap of garbage and gypsy camp
...
> stop at boulangerie for baguette enroute
> eur2 bread is better than anything in a 5-acre wal-mart
> office has no a/c, is next to heap of garbage and gypsy camp
...
1
0
1
29
@true_jae_yang > everyone in office has maths PhD and earns eur1200/month
> build superintelligence using minitel sets and renault 4
> get closed down by state
> workers move to london to bet on security prices
🇫🇷
> build superintelligence using minitel sets and renault 4
> get closed down by state
> workers move to london to bet on security prices
🇫🇷
1
0
1
47
@true_jae_yang churchill reached out FROM BEYOND THE GRAVE to make that man go to waterloo
what a hero
what a hero
0
0
1
23