4 ms·

I think you mean Google is following the leadership of Chainer, like Facebook already does? PyTorch started as a Chainer fork. Its dynamic graph internals are a

I think you mean Google is following the leadership of Chainer, like Facebook already does? PyTorch started as a Chainer fork. Its dynamic graph internals are all from Chainer.

bradleyjg 9y ago
This isn't art. There are no points for originality. If open source projects borrow the best parts from each other, that's a good thing.
- ychujo2 9y ago
  It's not a bad thing. It's good for users. But give credit to the leaders in the field. If you make an iPod clone, you call it an iPod clone, not a clone of the Zume HD. Chainer started it, was around years earlier, and it still has more users. So Google is not copying PyTorch, it's copying Chainer.
  - arbie 9y ago
    Do you have a source for Chainer currently having more users than TensorFlow?
    - Narew 9y ago
      Not more users than Tensorflow. But maybe more user than other dynamic deeplearning framework (PyTorch, Gluon, DyNet...)
- chrisprobert 9y ago
  Totally agree!
ma2rten 9y ago
This is the first time I am hearing this. I though pytorch was based on torch (like the name implies). Do you have a reference or more information?
- Narew 9y ago
  PyTorch use the same backend as torch (cutorch for GPU ...) But PyTorch use almost the same Python API than Chainer. On this point we can say PyTorch "copy" Chainer.