← All IntelClip / AI AgentsThe bet: under 1% of rollout-visible weights change per version
From Taking Reinforcement Learning Cross Datacenter — Nan Jiang, Modal · ≈6:03
States the core hypothesis precisely — only the served precision view of the weights matters, not FP32 master weights or optimizer moments — and the three-orders-of-magnitude payoff if it holds.
What’s in it
- States the core hypothesis precisely — only the served precision view of the weights matters, not FP32 master weights or optimizer moments — and the three-orders-of-magnitude payoff if it holds.
Clip transcript
synchronization. So the next question is can we keep the exact same serve version there but send a much smaller object. So this is the bet. What if less than 1% of rollout visible weights got changed from one version to another one by rollout visible weights I mean the weights in the served rollout checkpoint uh not the FP32 optimizer states not the atom like moments like the weights are the rollouts engine which will actually use to serve maybe let's say the FBA or maybe MVP4 format if that's true we do not need to shift the entire full parameter over the network we just need to shift the change the server view the precision data got different. The important part here is like still a bit wise reconstruction. The rollout engine gets the same served served version. You would have gotten to as sync to the full checkpoint there. So if this works then the link shrinks from hundreds of gigabytes to maybe hundreds of megabytes and this is something smaller enough we can just send it across like the
Comments
Checking sign-in…
Loading comments…