from Hacker News

PyTorch 1.0 is out

by kashifr on 12/7/18, 7:25 PM with 70 comments

  • by htfy96 on 12/7/18, 9:16 PM

    What surprised me most is the elegance of C++ API. Compared to its equivalence in Python, the C++ version is almost the same if we discard the "auto" keyword [0]. As mentioned in the doc, they put user-friendliness over micro-optimizations, which also proves the expressiveness of modern C++ (at least when they want to prioritize user-friendliness!) [0]: https://pytorch.org/cppdocs/frontend.html#end-to-end-example
  • by nafizh on 12/7/18, 10:55 PM

    Really grateful to the FAIR team for Pytorch. I use deep learning for computational biology. Pytorch lets me focus on the problem rather than nitpicking with the framework (looking at you tensorflow) to make something work.
  • by wpietri on 12/7/18, 10:04 PM

    In case anybody else was wondering, since this isn't in the fine article:

    "PyTorch is a Python package that provides two high-level features:

    * Tensor computation (like NumPy) with strong GPU acceleration

    * Deep neural networks built on a tape-based autograd system

    You can reuse your favorite Python packages such as NumPy, SciPy and Cython to extend PyTorch when needed."

  • by pesenti on 12/7/18, 11:42 PM

  • by tbenst on 12/7/18, 8:52 PM

    The new JIT is very interesting. Anyone know if this is for inference only or also for training?
  • by l3robot on 12/7/18, 9:34 PM

    TL;DR

    - New JIT feature that lets you run your model without python. It now seems trivial to load a pytorch model in C++

    - New distributed computation package. Major redesign.

    - C++ frontend

    - New torch hub feature to load models from github easily

  • by ritoune on 12/8/18, 2:08 AM

    I've been using PyTorch, and the PyTorch 1.0 pre-release for a while now. I adore it but don't really want to write C++ backends in production.

    Anyone want to start working on Golang bindings for C++ PyTorch?

  • by Rafuino on 12/7/18, 9:14 PM

    Is this the version combining Caffe2 and PyTorch into one framework? Is LMDB continuing as the default data load/store method?
  • by pilooch on 12/8/18, 8:46 AM

    If someone from fair is around, what does this mean for caffe2 ? Is the new torch c++ API a form a replacement for caffe?
  • by Solar19 on 12/8/18, 4:09 AM

    Agreed. I've always thought, however, that D would be the perfect language for a new Python implementation.
  • by dspoka on 12/7/18, 8:44 PM

    Is there some sort of pytorch 1.0 migration guide or does anyone know if there is any breaking from .41 to 1.0 ?
  • by openbasic on 12/7/18, 10:37 PM

    What would be a good book and project to get started with this? Object recognition? Product recommendations?
  • by dekimir on 12/8/18, 3:19 PM

    Hm, the Mac version of LibTorch is suddenly unavailable!? [0] I swear it was available for download until a few days ago...

    [0] https://pytorch.org/get-started/locally/

  • by amelius on 12/7/18, 8:46 PM

    > The JIT is a set of compiler tools for bridging the gap between research in PyTorch and production. It allows for the creation of models that can run without a dependency on the Python interpreter and which can be optimized more aggressively. Using program annotations existing models can be transformed into Torch Script, a subset of Python that PyTorch can run directly.

    Isn't python bytecode simple enough that it can be run anywhere?

  • by punnerud on 12/8/18, 11:53 AM

    Feels good to see torch-1.0.0 when I do: "python -m pip install torch torchvision --upgrade"