- 19 10月, 2016 40 次提交
-
-
由 Benoit Steiner 提交于
Change: 136597998
-
由 A. Unique TensorFlower 提交于
Change: 136596757
-
由 A. Unique TensorFlower 提交于
Change: 136566426
-
由 A. Unique TensorFlower 提交于
Change: 136565493
-
由 Justine Tunney 提交于
Many languages, e.g. JS, proto, etc. don't support having hyphens in package names. Change: 136565198
-
由 A. Unique TensorFlower 提交于
Change: 136560739
-
由 Justine Tunney 提交于
If I check out TensorFlow to a directory named tensorflow-clean then Bazel is going to create a subdirectory named bazel-tensorflow-clean. So we really need to have a wildcard pattern. Change: 136559940
-
由 A. Unique TensorFlower 提交于
Change: 136558601
-
由 A. Unique TensorFlower 提交于
This adds support for "NC*" data layouts for N-D convolution and pooling (including atrous convolution and pooling); previously, only "N*C" data lyaouts were supported. This also adds support for 1-D pooling (by forwarding to the 2-D implementation), and fixes the broken data_format support in conv1d. Change: 136556507
-
由 A. Unique TensorFlower 提交于
Change: 136556197
-
由 A. Unique TensorFlower 提交于
Change: 136555768
-
由 Justine Tunney 提交于
This includes ~200 -Wsign-compare warnings in TensorFlow, 15 -Wimplicit-function-declaration warnings in zlib, and more. I'm also working on getting rid of 400 protobuf warnings here: google/protobuf#2261 Change: 136555503
-
由 Saurabh Saxena 提交于
Change: 136554980
-
由 A. Unique TensorFlower 提交于
Change: 136553836
-
由 Derek Murray 提交于
This enables tf.contrib to be used on Windows, which does not currently support the TensorFlow plugin mechanism. Change: 136550845
-
由 A. Unique TensorFlower 提交于
Change: 136548976
-
由 Patrick Nguyen 提交于
Change: 136547976
-
由 A. Unique TensorFlower 提交于
Change: 136546018
-
由 A. Unique TensorFlower 提交于
Change: 136545648
-
由 Patrick Nguyen 提交于
Change: 136543444
-
由 A. Unique TensorFlower 提交于
Change: 136542181
-
由 Dan Mané 提交于
Also adds convenience method summary.get_summary_description. Change: 136542025
-
由 A. Unique TensorFlower 提交于
. grow the table when it is full. . add support for checkpointing and restoring. . allow vectors as values. . allow vectors as keys, including corresponding changes to common lookup code. . make sure the op is placed on parameter servers. . register the kernel for more data types. Change: 136541899
-
由 A. Unique TensorFlower 提交于
Change: 136541688
-
由 Patrick Nguyen 提交于
Change: 136541611
-
由 Dan Mané 提交于
When the EventAccumulator encounters a MetaGraphDef, it updates the _graph property. So there is no need to check when the Graph() method is called. Change: 136541160
-
由 Patrick Nguyen 提交于
Change: 136540437
-
由 Andrew Selle 提交于
We are holding off on switchover to the new ops for operator overloading and the Python API because of forward compatibility. - Implemented new FloorMod and FloorDiv which represent new implementations that match Pythonic semantics. These are not yet vectorized, but they are also not yet used. - Add TruncateMod in preparation to rename tf.mod to tf.truncateMod - Add RealDiv which only works on floating point inputs, but it uses Div's kernels - Add TruncateDiv which only works on integer inputs, but it also uses Div's kernels. Change: 136539473
-
由 A. Unique TensorFlower 提交于
Change: 136538420
-
由 A. Unique TensorFlower 提交于
Change: 136538067
-
由 A. Unique TensorFlower 提交于
Change: 136533159
-
由 A. Unique TensorFlower 提交于
Change: 136528360
-
由 A. Unique TensorFlower 提交于
Change: 136527767
-
由 A. Unique TensorFlower 提交于
Change: 136524312
-
由 A. Unique TensorFlower 提交于
Change: 136522712
-
由 A. Unique TensorFlower 提交于
value is written but never read. Also disable svd_op_test in msan - the error is reported from within numpy. Change: 136521729
-
由 Sukriti Ramesh 提交于
Change: 136521066
-
由 A. Unique TensorFlower 提交于
Change: 136520720
-
由 Sukriti Ramesh 提交于
Change: 136520082
-
由 Rohan Jain 提交于
Adding an atomic_write method to file_io which uses a temp file and renaming to "atomically" write to a file. Fixes this in training_util.py which is used by saver.py Change: 136519920
-