My Assessment of Compute Verification in Plan A (+ open questions)

·LessWrong··

OverviewThese are my non-expert notes on the compute verification section of AIFP’s Plan A. I cover interconnect limits, memory wipes, network taps + replay, and ZKPs. For the most part, the sections can be read independently. I restrict my attention to inference-only verification: ensuring that compute is used for inference, not training. For each method suggested by AIFP, I ask:How much can it slow down training?How much overhead does it add to inference?What sensitive information does it requ...

Read full article →

Related Articles

GCC steering committee announces AI policy
arto · Hacker News · 11h ago
Why is everyone trying to build a solid-state battery?
crescit_eundo · Hacker News · 10h ago
Stacked PRs are now live on GitHub
tomzorz · Hacker News · 6h ago
AI's top startups are barely publishing their research
YeGoblynQueenne · Hacker News · 1d ago
Why DNA damage from smoking and UV rays cause cancer in some but not others
gmays · Hacker News · 8h ago