AdamW and Super-convergence is now the fastest way to train neural nets
241 by tim_sw | 26 comments on Hacker News.
Wednesday, July 4, 2018
New top story on Hacker News: AdamW and Super-convergence is now the fastest way to train neural nets
Related Posts:
New top story on Hacker News: Certificate Transparency Log Compromised via Salt Vulnerability- New top story on Hacker News: Comparing Software Engineer Salaries Across Cities
- New top story on Hacker News: What Makes a Good Teacher?
- New top story on Hacker News: Windows 10 has fake “Python” command that redirects to the Windows Store
- New top story on Hacker News: Shorten the Workweek to Reopen Safely
0 comments:
Post a Comment