X · @teortaxesTex
· X / Twitter
> the real limitation being RL flops right now I wonder what the limits of parallelization with MOPD are obviously, 1000 "experts" with 20 RL steps ea…
> the real limitation being RL flops right nowI wonder what the limits of parallelization with MOPD areobviously, 1000 "experts" with 20 RL steps each are ≈useless compared to 10 experts with 2000 steps. But what about 40 experts @ 500 steps, merging@4 before OPD?xjdr: around feb / starting with gpt 5.2, model capabili