New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Clarify batch_norm documentation to highlight that the dimensions for… #15653
Clarify batch_norm documentation to highlight that the dimensions for… #15653
Conversation
Branch 178996911
* Random flip benchmark * More benchmarks
- Use `nvidia/cuda:9.0-base-ubuntu16.04` as the base image to select just the CUDA libraries we need. - Remove the installed static libraries. - Remove the dependency on openjdk-8 since Bazel ships with a local copy. - Perform a shallow clone of the repository. The image is 2.94GB, down from 4.87GB. Signed-off-by: Felix Abecassis <fabecassis@nvidia.com>
Fix issue in the `Defun` docs
CMake: configure default string values of options properly
Fixed typos in comments
Forward declare condition_variable, fix for #14388
Change "'", which causes incorrect highlight, to "`"
- Adding comment about error if not casted to long first - Return error message if resolving path failed on FreeBSD
* Don't add -ltensorflow_framework to sysconfig LFLAGS if the build is monolithic
…file index_to_string_table_from_file can use tf.string as vocabulary file
Add description for new PR workflow.
This fix is an effort to add libsvm format support with the implementation of `decode_libsvm`, as was proposed in 14313. The implementation is done in contrib with: This fix fixes 14313. Signed-off-by: Yong Tang <yong.tang.github@outlook.com>
Signed-off-by: Yong Tang <yong.tang.github@outlook.com>
Signed-off-by: Yong Tang <yong.tang.github@outlook.com>
* Add templated functions for `safe_strto[f|d|i32|i64|u32|u64]` While working on 14330 I noticed that there is no templated functions for `safe_strto[f|d|i32|i64|u32|u64]`. As a result, an additional wrapper has to be created in different places to apply the typename in a templated class or function. Examples include the existing implementations in `tensorflow/core/kernels/string_to_number_op.cc` (`Convert`), `tensorflow/core/lib/strings/proto_text_util.h` (`ProtoParseNumeric`), and in 14330. It might make sense to add a templated function for `safe_strto[f|d|i32|i64|u32|u64]` to avoid existing and future code duplications. This fix adds ``` template<typename T> bool SafeStringToNumeric(StringPiece s, T* value); ``` to address the above mentioned issue. Signed-off-by: Yong Tang <yong.tang.github@outlook.com> * Move ProtoParseNumeric to numbers.h, and use overloading in template Signed-off-by: Yong Tang <yong.tang.github@outlook.com>
* Add int64 support for BroadcastArgs and BroadcastGradientArgs In `array_ops.cc`, both int32 and int64 are expected to be supported for `BroadcastArgs` and `BroadcastGradientArgs`. However, this was not the case as only int32 kernel are registered even though `T` is part of the `TypeConstraint`. This fix adds the int64 kernel support for `BroadcastArgs` and `BroadcastGradientArgs`, and adds related test cases. Signed-off-by: Yong Tang <yong.tang.github@outlook.com> * Add test cases for int64 support of BroadcastArgs and BroadcastGradientArgs Signed-off-by: Yong Tang <yong.tang.github@outlook.com> * Sanitize python and clang-format. Signed-off-by: Yong Tang <yong.tang.github@outlook.com>
* Fix discrepancy between doc and impl for `tf.reverse` This fix tries to address the discrepancy between doc and implementations for `tf.reverse`. In the documentation, both int32 and int64 could be used for axis. However, the actual implementation of the `tf.reverse` does not support int64 and caused missing kernel error: ```python ubuntu@ubuntu:~$ python Python 2.7.12 (default, Nov 19 2016, 06:48:10) [GCC 5.4.0 20160609] on linux2 Type "help", "copyright", "credits" or "license" for more information. >>> import tensorflow as tf >>> v = tf.reverse_v2([1, 2, 3], tf.constant([0], tf.int64)) >>> sess = tf.Session() >>> sess.run(v) Traceback (most recent call last): File "<stdin>", line 1, in <module> File "/usr/local/lib/python2.7/dist-packages/tensorflow/python/client/session.py", line 889, in run run_metadata_ptr) File "/usr/local/lib/python2.7/dist-packages/tensorflow/python/client/session.py", line 1120, in _run feed_dict_tensor, options, run_metadata) File "/usr/local/lib/python2.7/dist-packages/tensorflow/python/client/session.py", line 1317, in _do_run options, run_metadata) File "/usr/local/lib/python2.7/dist-packages/tensorflow/python/client/session.py", line 1336, in _do_call raise type(e)(node_def, op, message) tensorflow.python.framework.errors_impl.InvalidArgumentError: No OpKernel was registered to support Op 'ReverseV2' with these attrs. Registered devices: [CPU], Registered kernels: device='CPU'; T in [DT_STRING]; Tidx in [DT_INT32] device='CPU'; T in [DT_BOOL]; Tidx in [DT_INT32] device='CPU'; T in [DT_COMPLEX128]; Tidx in [DT_INT32] device='CPU'; T in [DT_COMPLEX64]; Tidx in [DT_INT32] device='CPU'; T in [DT_DOUBLE]; Tidx in [DT_INT32] device='CPU'; T in [DT_FLOAT]; Tidx in [DT_INT32] device='CPU'; T in [DT_HALF]; Tidx in [DT_INT32] device='CPU'; T in [DT_INT8]; Tidx in [DT_INT32] device='CPU'; T in [DT_UINT8]; Tidx in [DT_INT32] device='CPU'; T in [DT_INT16]; Tidx in [DT_INT32] device='CPU'; T in [DT_UINT16]; Tidx in [DT_INT32] device='CPU'; T in [DT_INT32]; Tidx in [DT_INT32] device='CPU'; T in [DT_INT64]; Tidx in [DT_INT32] [[Node: ReverseV2 = ReverseV2[T=DT_INT32, Tidx=DT_INT64](ReverseV2/tensor, Const)]] ``` This fix addresses this issue and added the int64 support for axis in `tf.reverse`. Signed-off-by: Yong Tang <yong.tang.github@outlook.com> * Add GPU support of int64 axis in `tf.reverse` Signed-off-by: Yong Tang <yong.tang.github@outlook.com> * Add test cases for int64 support of axis in `tf.reverse` Signed-off-by: Yong Tang <yong.tang.github@outlook.com> * Add OpenCL int64 support of axis in `tf.reverse` Signed-off-by: Yong Tang <yong.tang.github@outlook.com>
* ENH: implement cohen_kappa * TST: add test cases * DOC: add document * DOC: revise document as suggested * CLN: drop total prefix * CLN: wrap lines to 80 chars in metric_ops.py * CLN: remove unneeded dependency * DOC: support labels datatype * TST: add explanation for testBasic * TST: remove testBasic2 * TST: replace all random data * CLN: remove sklearn dependency * TST: wrap the lines in this test to 80 chars * DOC: clean format * CLN: typo * CLN: lables specifiy -1 axis * CLN: remove float lables support * DOC: modify document of _safe_div * CLN: rename predictions -> predictions_idx * TST: invalid dim case * CLN: clean codes * DOC: remove total_ in update_op
* mpi_collectives: Refactor to fix build issues After TF commit 5c7f9e3, the mpi_collectives package would no longer build ops and kernels. This build issue caused mpi_collectives import to fail in Python with the following error: "NameError: Could not find operator MPISize in dynamic library mpi_collectives.so". To fix this issue, add build targets to ensure both ops and kernels are built Note, also refactored the build targets and directory structure to more closely match other contrib packages. * mpi_collectives: Minor BUILD fix * Minor: Buildifier fix * mpi_collectives: Correct preprocessor defines * mpi_collectives: Clearer defines inclusion
…format (#14120) * TST: add test case * BUG: fix unknown spatial dim for NC format * Make comment more informative
…tracking_alloc_test fixed for GPU (#14530) * session_clusterspec_prop_test fix for GPU * common_runtime_direct_session_with_tracking_alloc_test fix for GPU
… normalization depend on the shape of the tensor.
… normalization depend on the shape of the tensor.
Can one of the admins verify this patch? |
Thanks for your pull request. It looks like this may be your first contribution to a Google open source project. Before we can look at your pull request, you'll need to sign a Contributor License Agreement (CLA). 📝 Please visit https://cla.developers.google.com/ to sign. Once you've signed, please reply here (e.g.
|
I signed it! |
We found a Contributor License Agreement for you (the sender of this pull request), but were unable to find agreements for the commit author(s). If you authored these, maybe you used a different email address in the git commits than was used to sign the CLA (login here to double check)? If these were authored by someone else, then they will need to sign a CLA as well, and confirm that they're okay with these being contributed to Google. |
Yes I used a different e-mail address for the commit. I cannot add this address to the google account where I signed the contract. |
Could you change the email address on the commit? |
Ok I changed the e-mail address in my commit. However this is not a standard procedure with git and requires to rewrite all commits. As far as I can see, this pull-request just contains my change, so I think it should work now. |
OK, maybe let's wait until the next refresh. |
@Netzeband Still broken. I don'y anything pushed. You can amend the commit with |
…/Netzeband/tensorflow into clarify_batch_norm_documentation
I think this will go into the wrong direction, I will close the request and setup a new one, where not the whole commit history have been rewritten (due to the e-mail address change). |
I created a new request: #15728 |
... normalization depend on the shape of the tensor.