crocomama

↗ HN profile [ 88.7 ms ] full profile
Karma
1
Created
December 21, 2023 (2y ago)
Submissions
0
I like ML and Systems stuff.
  1. Instead of running your neural networks either fully on cloud or fully on edge, why not split it between the two? Saves the cost of transporting all the weights to the edge, and saves cloud gpu inference costs, since…