by

Where communities thrive


  • Join over 1.5M+ people
  • Join over 100K+ communities
  • Free without limits
  • Create your own community
People
Activity
    Sameroom
    @sameroom-bot
    [Seiya Tokui, chainer] So you can emulate large batch training by multiple backprops with small batchsize.
    Sameroom
    @sameroom-bot
    [Satoshi Tsutsui, chainer] Wow! Thanks for the reply! That's great to know!
    Sameroom
    @sameroom-bot
    [dhgrs, chainer] If the network includes batch normalization or other normalization, it does not work well, right?
    Sameroom
    @sameroom-bot
    [Ishan Rai, chainer] I am getting the following error while running test after recently pulled from the master. The tests are successful on older branches:
    [Ishan Rai, chainer] What could I do to resolve it?
    Sameroom
    @sameroom-bot
    [Seiya Tokui, chainer] Let me check
    [Seiya Tokui, chainer] Did you recompile after the pull?
    Sameroom
    @sameroom-bot
    [Ishan Rai, chainer] No, I did not
    Sameroom
    @sameroom-bot
    [Ishan Rai, chainer] error.gif
    [Ishan Rai, chainer] Take a look at this gif ^^
    Sameroom
    @sameroom-bot

    [Ishan Rai, chainer] (env)ishan@Wall-E ~ $ git reset --hard HEAD^ HEAD is now at 5cdcdc4 Merge pull request #6204 from okapies/feature/cupy-to-chx (env)ishan@Wall-E ~ $ python -m pytest test_flip.py ImportError while loading conftest '/home/ishan/Desktop/openSource/VChainer/src/chainer/tests/chainer_tests/conftest.py'. ../../conftest.py:3: in <module> from chainer.backends import cuda ../../../../chainer/__init__.py:10: in <module> from chainer import backends # NOQA ../../../../chainer/backends/__init__.py:1: in <module> from chainer.backends import cuda # NOQA ../../../../chainer/backends/cuda.py:45: in <module> from chainer.backends import _cpu ../../../../chainer/backends/_cpu.py:8: in <module> import chainerx ../../../../chainerx/__init__.py:23: in <module> from chainerx._core import _to_cupy # NOQA E ImportError: cannot import name '_to_cupy' (env)ishan@Wall-E ~ $ git reset --hard HEAD^ HEAD is now at abe80b4 Merge pull request #6941 from hvy/align-macro-arg (env)ishan@Wall-E ~ $ python -m pytest test_flip.py ^[[AImportError while loading conftest '/home/ishan/Desktop/openSource/VChainer/src/chainer/tests/chainer_tests/conftest.py'. ../../conftest.py:3: in <module> from chainer.backends import cuda ../../../../chainer/__init__.py:10: in <module> from chainer import backends # NOQA ../../../../chainer/backends/__init__.py:1: in <module> from chainer.backends import cuda # NOQA ../../../../chainer/backends/cuda.py:45: in <module> from chainer.backends import _cpu ../../../../chainer/backends/_cpu.py:8: in <module> import chainerx ../../../../chainerx/__init__.py:55: in <module> _docs.set_docs() ../../../../chainerx/_docs/__init__.py:24: in set_docs m.set_docs() ../../../../chainerx/_docs/array.py:275: in set_docs ndarray.min, E AttributeError: type object 'chainerx.ndarray' has no attribute 'min'
    (env)ishan@Wall-E ~ $ git reset --hard HEAD^
    HEAD is now at c94a1c3 Merge pull request #6752 from Harshan01/amin-chainerx-new
    (env)ishan@Wall-E ~ $ python -m pytest test_flip.py
    ImportError while loading conftest '/home/ishan/Desktop/openSource/VChainer/src/chainer/tests/chainer_tests/conftest.py'.
    ../../conftest.py:3: in <module>
    from chainer.backends import cuda
    ../../../../chainer/__init_.py:10: in <module>
    from chainer import backends # NOQA
    ../../../../chainer/backends/
    _init__.py:1: in <module>
    from chainer.backends import cuda # NOQA
    ../../../../chainer/backends/cuda.py:45: in <module>
    from chainer.backends import _cpu
    ../../../../chainer/backends/_cpu.py:8: in <module>
    import chainerx
    ../../../../chainerx/__init__.py:55: in <module>
    _docs.set_docs()
    ../../../../chainerx/_docs/__init__.py:24: in set_docs
    m.set_docs()
    ../../../../chainerx/_docs/array.py:275: in set_docs
    ndarray.min,
    E AttributeError: type object 'chainerx.ndarray' has no attribute 'min'
    (env)ishan@Wall-E ~ $ git reset --hard HEAD^
    HEAD is now at e26591e Merge pull request #6845 from asi1024/fp16tol
    (env)ishan@Wall-E ~ $ python -m pytest test_flip.py
    ========================================================================================================= test session starts ==========================================================================================================
    platform linux -- Python 3.6.5, pytest-4.3.0, py-1.7.0, pluggy-0.8.1
    rootdir: /home/ishan/Desktop/openSource/VChainer/src/chainer, inifile: setup.cfg
    collected 152 items

    test_flip.py ........................................................................................................................................................ [100%]
    `\

    [Kenichi Maehashi, chainer] You need to recompile after git pull.
    Sameroom
    @sameroom-bot
    [Ishan Rai, chainer] Do I have to recompile after every git pull?
    [Ishan Rai, chainer] Because after I never recompiled previously and the there were no error. Also I tracked down the commits that are causing error ^^
    Sameroom
    @sameroom-bot
    [Masaki Kozuki, chainer] Yes we have to compile after every git pull.
    Sameroom
    @sameroom-bot
    [Ishan Rai, chainer] Thanks, I will do that.
    Sameroom
    @sameroom-bot
    [Ishan Rai, chainer] I recompiled and there is no error now. :)
    Sameroom
    @sameroom-bot
    [Aniket Didolkar, chainer] When I run scripts/run-clang-format.sh it uses a default configuration rather than .clang-format in chainerx_cc?How do I fix this?
    Sameroom
    @sameroom-bot
    [Crissman Loomis, chainer]
    here With the abdication of the Japanese emperor and beginning of a new era, Japan will have a long Golden Week holiday week, beginning tomorrow until May 7.
    During this time, many of the Chainer maintainers will be on vacation. Most responses will be after Golden Week.
    Thank you for your participation in Chainer!
    Sameroom
    @sameroom-bot
    [Masaki Kozuki, chainer] Is there any plan to support Ubuntu 18.04 as a recommended environment of Chainer/CuPy?
    Sameroom
    @sameroom-bot
    [Wang Yu, chainer] Hello, Is there anyone know how can I send a model which has a ‘list’ attribute such as ‘self.ln = [ L.linear([3,5]) ]’, to GPU?
    because after I used ‘model.to_gpu’, the variables of linear are still numpy.narray.:)
    Sameroom
    @sameroom-bot
    [nishino, chainer] You have to override Link.to_gpu in your model to transfer the linear manually (by writing something like for l in self.ln: l.to_gpu()).
    From v6, even though the above method should work, the recommended way will be to override Link.device_resident_accept instead. It's not documented yet but the implementation should be something like
    def device_resident_accept(self, visitor):
        super().device_resident_accept(visitor)
        for l in self.ln:
            l.device_resident_accept(visitor)
    Sameroom
    @sameroom-bot
    [Wang Yu, chainer] I see. Thank you very much
    Sameroom
    @sameroom-bot
    [Wang Yu, chainer] @nishino Hello Mr. nishino. I also found that both ‘optimizer.update()’ and ‘model.cleargrads’ can not clear the grads if the model has a ‘list’ attribute such as ‘self.ln = [ L.linear([3,5]),]’. I have to clear the grads one by one by using for-loop. In this case, is there any better way?(sweat smile emoji)
    [Masaki Kozuki, chainer] I don’t understand the context enough, but how about ChainList?
    https://docs.chainer.org/en/stable/reference/generated/chainer.ChainList.html#chainer.ChainList
    Sameroom
    @sameroom-bot
    [Wang Yu, chainer] (ohcrap emoji)It’s useful for me. I’ll try. thank you @crcrpar
    Sameroom
    @sameroom-bot
    [mingxiao huang, chainer] @jing.xu
    Sameroom
    @sameroom-bot
    [Crissman Loomis, chainer] For our test rewrite tracking issue #6071: 76 done of 165 tests, 46% complete!
    Sameroom
    @sameroom-bot
    [Akshay Subramanian, chainer] Untitled
    Sameroom
    @sameroom-bot
    [nishino, chainer] It seems chainerx is not built correctly. Could you check chainerx.is_available()? I guess it returns False.
    Sameroom
    @sameroom-bot
    [Akshay Subramanian, chainer] Yes. It does return False. Will try rebuilding. Thanks for the suggestion!
    Sameroom
    @sameroom-bot
    [Ace, chainer] Can anyone explain the "relationship" among array, array_node and array_body in chainerx :D thank you. (especially array_node)
    Sameroom
    @sameroom-bot

    [Masaki Kozuki, chainer] What I write here might be a bit obsolete, but according to chainer/chainer#6423,

    chainerx::ArrayNode: A node representing an array in the computational graph. It is owned by an chainerx::ArrayBody.

    Also, https://github.com/chainer/chainer/blob/master/chainerx_cc/chainerx/array.h#L46 says Array wraps an ArrayBody.

    Sameroom
    @sameroom-bot
    [Ace, chainer] Thank you, I find the https://www.slideshare.net/HiroyukiVincentYamaz/chainerx-and-how-to-take-part is also a good introduction.
    Sameroom
    @sameroom-bot
    [yixuan, chainer] hello all, one question about Chainer + CUDA 10, does it support CUDA 10?which version?
    Sameroom
    @sameroom-bot
    [Masaki Kozuki, chainer] I think chainer supports CUDA 10 since v5.2 and the current stable (v6.0.0) supports CUDA 10.1 also.
    Sameroom
    @sameroom-bot
    [yixuan, chainer] @crcrpar does there have any guide to build chaine template with cuda 10?
    Sameroom
    @sameroom-bot

    [Masaki Kozuki, chainer] The installation guides of Chainer & CuPy are enough I think.

    For cuda10.0 environment, I think the below is enough according to the above guides.

    $ pip install -U setuptools pip
    $ pip install cupy-cuda100
    $ pip install chainer
    Sameroom
    @sameroom-bot
    [yixuan, chainer] thanks for your information. i will try later @crcrpar
    Sameroom
    @sameroom-bot
    [Akshay Subramanian, chainer] Untitled
    [Akshay Subramanian, chainer] chainerx.is_available() is also returning False.
    Sameroom
    @sameroom-bot
    [Kshiteej K, chainer] I faced this issue few times when, there was a mismatch in function declaration and definition. Maybe check the code once to see that is correct.
    Sameroom
    @sameroom-bot
    [Wang Yu, chainer] Hello, Beg your help, please. I just want to calculate the cross_entropy between two variables without using softmax, so is there any function available?
    Sameroom
    @sameroom-bot
    [Wang Yu, chainer] Thank you for your answer, but I just want the cross_entropy without any activation function.
    Sameroom
    @sameroom-bot
    [Masaki Kozuki, chainer] OK, then I would implement that function using chainer.functions. :)
    Sameroom
    @sameroom-bot
    [Wang Yu, chainer] Thanks a lot:D
    Sameroom
    @sameroom-bot
    [Anshuk Uppal, chainer] Sorry for troubling you guys with this but I couldn't find anything definite upon googling. Q: Does Chainer support backprop on vector losses? I have been told so before but simple codes do not work, am I missing something here?
    Sameroom
    @sameroom-bot
    [Anshuk Uppal, chainer] As much as I could trace the problem the grad_var being passed in line 1438(variable.py) : _backprop_to_all([(node, grad_var)], retain_grad, loss_scale) is None and this propagates through to the gradient calculation
    Sameroom
    @sameroom-bot
    [Katsuhiko Nishimra, chainer] Why not F.sum(vector).backward()?