from Hacker News

PyTorch: An Imperative Style, High-Performance Deep Learning Library [pdf]

by stablemap on 12/6/19, 5:08 PM with 16 comments

• by jspisak on 12/6/19, 6:59 PM

  Finally the solution to all of your PyTorch citation problems! :)

• by scythe on 12/6/19, 11:17 PM

  Man it’s kinda sad as a Lua fan to see so much interest in a project where the main goal is just to not use Lua.

  I guess academics like familiarity and Lua insistently refuses to be like other languages (arrays and maps in one type, 1-based arrays, nonstandard builtin patterns, etc).

• by amrrs on 12/6/19, 7:25 PM

  If you want to read it online - https://www.arxiv-vanity.com/papers/1912.01703/

• by zapnuk on 12/7/19, 12:27 AM

  Is there a typo in Listing 1?

  The forward function of the conv net should use:

  t3 = self.fc(t2)

  instead of:

  t3 = self.fc(t1)

  AFAIK the nn.functional.relu function is NOT inplace by default [1]

  https://pytorch.org/docs/stable/nn.functional.html

• by foxes on 12/6/19, 11:18 PM

  It's a bit funny to call it imperative, when really at the end of the day, the objective is to get something where you have very little insight into what the neural net. is doing to the state.