HN Explorer

gyang

↗ HN profile [ 60.2 ms ] full profile
Karma
127
Created
April 22, 2017 (9y ago)
Submissions
0
  1. Tuning GPT-3 on a single GPU (twitter.com)
    4 points 0
  2. On infinitely wide neural networks that exhibit feature learning (microsoft.com)
    3 points 0
  3. Wide Neural Networks Are Feature Learners, Not Kernel Machines (arxiv.org)
    5 points 0
  4. Validation loss is not a reliable signal for early stopping (github.com)
    1 points 0
  5. In ML, when loss decreases, accuracy increases, right? (twitter.com)
    1 points 0
  6. Provably Robust Deep Learning (decentdescent.org)
    1 points 0
  7. Wide Neural Networks of Any Architecture Are Gaussian Processes (arxiv.org)
    144 points 35
  8. Lie-Access Neural Turing Machines: Manifolds as Memory Storage (lstm.seas.harvard.edu)
    10 points 1