| Commit message (Collapse) | Author | Age |
|
|
|
| |
Change: 140883595
|
|
|
|
| |
Change: 140880753
|
|
|
|
| |
Change: 140878296
|
|
|
|
|
|
| |
tf.strided_slice in the following 15 files. more to come!
Change: 140875384
|
|
|
|
| |
Change: 140871440
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
objects.
See the added javadoc comments and unittests (TensorTest.java) for
an explanation of the API. The conversion between Java types and the
Tensor class does involve a single copy.
For Java object -> TF_Tensor, a single JNI function (setValue) suffices.
For TF_Tensor -> Java object, I couldn't work out a scheme that didn't
require a separate JNI function for each primitive type and one JNI function
for non-scalars.
However, since all these JNI methods are private and not part of the
exposed Java API, I figured the "cost" asymmetry in the conversion functions
was worth the "benefit" of fewer JNI methods.
Another step towards #5
Change: 140871139
|
|
|
|
| |
Change: 140870764
|
|
|
|
|
| |
Internal BUILD visibility change
Change: 140869738
|
|
|
|
| |
Change: 140864110
|
|
|
|
| |
Change: 140860357
|
|
|
|
| |
Change: 140860022
|
|
|
|
| |
Change: 140859776
|
|
|
|
| |
Change: 140858933
|
|
|
|
| |
Change: 140857009
|
|
|
|
| |
Change: 140855283
|
|
|
|
|
|
|
|
|
|
| |
Previously, the predict API of the Evaluator, when called with only
non-deprecated arguments, always returns a never-ending generator.
This is counterintuitive and runs against our public documentation.
This change modifies the predict API to only run through the input
examples once, if there are no queue-runners and therefore the second
epoch will be identical to the first.
Change: 140854520
|
|
|
|
| |
Change: 140854331
|
|
|
|
|
|
|
| |
dont know both of x,y. I.e., if there's a direct implementation of the {forward,inverse}_log_det_jacobian then we won't bother computing the x or y.
Also made several style fixes.
Change: 140851970
|
|
|
|
|
|
| |
expressions.
Change: 140845713
|
|
|
|
| |
Change: 140820143
|
|
|
|
| |
Change: 140819761
|
|
|
|
| |
Change: 140819261
|
|
|
|
|
|
| |
learn_io package.
Change: 140815606
|
|
|
|
| |
Change: 140814270
|
|
|
|
| |
Change: 140813882
|
|
|
|
| |
Change: 140813226
|
|
|
|
| |
Change: 140811785
|
|
|
|
| |
Change: 140808119
|
|
|
|
|
|
| |
--config=cuda.
Change: 140807677
|
|
|
|
| |
Change: 140807620
|
|
|
|
| |
Change: 140798911
|
|
|
|
| |
Change: 140797107
|
|
|
|
| |
Change: 140796727
|
|
|
|
| |
Change: 140794371
|
|
|
|
| |
Change: 140794278
|
|
|
|
| |
Change: 140793810
|
|
|
|
|
|
| |
contrib to core.
Change: 140793359
|
|
|
|
|
|
|
|
|
|
|
| |
tf.histogram_summary, tf.scalar_summary, tf.audio_summary, tf.image_summary, tf.merge_all_summaries, tf.merge_summary are all removed.
Nearly-identical and fully supported apis are available at
tf.summary.histogram, tf.summary.scalar, tf.summary.audio, tf.summary.image, tf.summary.merge_all, tf.summary.merge
The major change in the new API is that the summary "tag" is now actually the node name, which means the TF naming system will automatically deduplicate summary tags.
If you need an exact match for the old API, you may use tf.contrib.deprecated.histogram_summary, etc, but these endpoints will eventually be removed.
Change: 140792244
|
|
|
|
|
|
|
| |
tf.summary.merge_all
The implementations are identical, tf.summary.merge_all is the canonical location.
Change: 140790809
|
|
|
|
| |
Change: 140789077
|
|
|
|
|
| |
of specific variables to save or restore.
Change: 140788798
|
|
|
|
| |
Change: 140788482
|
|
|
|
|
|
|
|
|
|
|
| |
... wherein the flawed protagonist adds temporary support for *both* versions
of the property so that a followup change can perform the rename.
A final change will then remove the old `shape`.
Requires adding an iterator for SparseTensorValue because some folks use its
tuple iterable property.
Change: 140784809
|
|
|
|
| |
Change: 140784570
|
|
|
|
| |
Change: 140783483
|
|
|
|
|
|
| |
arrays as input function.
Change: 140782176
|
|
|
|
| |
Change: 140778111
|
|
|
|
| |
Change: 140778005
|
|
|
|
| |
Change: 140777702
|
|
|
|
| |
Change: 140777101
|