| Commit message (Collapse) | Author | Age |
... | |
|
|
|
| |
Change: 140776069
|
|
|
|
| |
Change: 140775287
|
|
|
|
| |
Change: 140773742
|
|
|
|
| |
Change: 140771346
|
|
|
|
|
|
| |
linear_operator_util.py added as well
Use linear_operator_util in LinearOperatorDiag, and a couple bug-fixes.
Change: 140770818
|
|
|
|
| |
Change: 140769904
|
|
|
|
| |
Change: 140764844
|
|
|
|
| |
Change: 140763104
|
|
|
|
|
| |
These classes are now available as tf.summary.FileWriter and tf.summary.FileWriterCache.
Change: 140762782
|
|
|
|
| |
Change: 140761568
|
|
|
|
| |
Change: 140761549
|
|
|
|
| |
Change: 140760957
|
|
|
|
|
|
|
|
|
|
|
|
| |
The current implementation has limitations, in particular
it does not support string tensors. But the same API
will be sufficient when that limitation has been
addressed.
See also #6003. The API added here can be used to
fill in the tensor_contents field of a TensorProto
protocol buffer.
Change: 140760816
|
|
|
|
|
|
| |
exactly the same as pack and unpack and will replace pack and unpack going
forward.
Change: 140760260
|
|
|
|
| |
Change: 140759688
|
|
|
|
|
|
| |
package easier.
Change: 140758957
|
|
|
|
| |
Change: 140754388
|
|
|
|
| |
Change: 140751451
|
|
|
|
|
|
| |
Remove the label accessor, replace with a string. The projector adapter knows how to reach into the dataset, which is ok for now.
Change: 140748823
|
|
|
|
| |
Change: 140748366
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
lower than both CPU and GPU, to prevent accidentally placing
ops on devices.
Most ops are implemented for either CPU and GPU, and this default can
slow down models that do not carefully place all of their ops
explicitly.
RELNOTES: non-CPU/GPU devices are no longer automatically preferred
over CPU/GPU devices: they must be explicitly named in the nodes to be
used.
Change: 140748128
|
|
|
|
| |
Change: 140744288
|
|
|
|
|
|
| |
Remove extra list used for android builds, and use the new list in core:BUILD
instead.
Change: 140743365
|
|
|
|
| |
Change: 140742376
|
|
|
|
| |
Change: 140740851
|
|
|
|
|
|
| |
tf.contrib.rnn package and moves the rnn_test.py as contrib core_rnn_test.py.
Change: 140739860
|
|
|
|
|
|
|
|
|
| |
(If the number of entries in the proto was zero, but we asked for
more, the code incorrectly dereferenced a pointer to the data before
the check to see if it was empty, resulting in a crash.)
Bug spotted by libFuzzer.
Change: 140736570
|
|
|
|
| |
Change: 140734296
|
|
|
|
| |
Change: 140733125
|
|
|
|
| |
Change: 140684634
|
|
|
|
| |
Change: 140682786
|
|
|
|
| |
Change: 140680070
|
|
|
|
| |
Change: 140678566
|
|
|
|
| |
Change: 140678361
|
|
|
|
| |
Change: 140678314
|
|
|
|
| |
Change: 140676866
|
|
|
|
| |
Change: 140676173
|
|
|
|
|
|
| |
Create temporary dummy arguments with old names at the end in preparation for
deprecation.
Change: 140675930
|
|
|
|
| |
Change: 140674798
|
|
|
|
| |
Change: 140674321
|
|
|
|
| |
Change: 140672917
|
|
|
|
|
|
|
| |
Allows declaring two Example keys (an index key and a value key) for a single
Feature, the returned value is a SparseTensor with indices given by the index
key.
Change: 140671682
|
|
|
|
| |
Change: 140670395
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
default value.
Eliminate duplicate bias centering of logits.
Merge duplicate training_loss and train_op code between different head classes. There's still duplicate code in head_op, but at least it's down to just one method now.
Pass `labels` tensor, not dict, to train_op.
Add name to predictions tensors.
Change non-mutable lists to tuples.
Remove unused `name` args.
Change: 140669937
|
|
|
|
| |
Change: 140667141
|
|
|
|
| |
Change: 140661470
|
|
|
|
|
|
| |
subclassing of Distribution.
Change: 140659334
|
|
|
|
|
|
|
|
|
|
| |
an ambiguous way to designate (1) the deps attribute of each generated
tf_kernel_library target, and (2) associated tf_kernel_library targets for ops
which should be included as dependencies of the omnibus cc_library target
(for example, :dilation_ops.)
As a result, each individual op library brought in the deps of all the others
along with dependencies on unrelated ops.
Change: 140657643
|
|
|
|
|
| |
- Update saved_model_half_plus_two to export a prediction default serving signature.
Change: 140657101
|
|
|
|
| |
Change: 140651442
|