| .. |
|
__init__.py
|
TensorFlow: Upstream a batch of changes to git.
|
2015-11-09 18:42:53 -08:00 |
|
accumulate_n_benchmark.py
|
Change accumulate_n() to no longer require the output shape to be known in advance, and to no longer blocked until the first input is available.
|
2016-08-08 12:18:13 -07:00 |
|
array_grad.py
|
Update trace, matrix_set_diag, matrix_diag_part and their gradients to work for rectangular matrices.
|
2016-09-29 13:35:30 -07:00 |
|
array_ops.py
|
Documentation fix: markdown requires an empty line before ``` block.
|
2016-09-29 12:26:21 -07:00 |
|
batch_norm_benchmark.py
|
Update copyright for 3p/tf/python.
|
2016-06-02 13:38:16 -07:00 |
|
candidate_sampling_ops.py
|
Delegate to C++ shape functions for various ops in candidate_sampling, ctc, io,
|
2016-09-07 11:03:12 -07:00 |
|
check_ops.py
|
Quote tf.* identifiers in documentation.
|
2016-09-29 00:50:02 -07:00 |
|
clip_ops.py
|
Merge changes from github.
|
2016-08-26 14:01:27 -07:00 |
|
control_flow_grad.py
|
Defer the check on second order gradient to the right place.
|
2016-09-15 09:46:04 -07:00 |
|
control_flow_ops_test.py
|
Added to_proto and from_proto for CondContext and WhileContext.
|
2016-09-13 10:17:12 -07:00 |
|
control_flow_ops.py
|
Quote tf.* identifiers in documentation.
|
2016-09-29 00:50:02 -07:00 |
|
ctc_ops.py
|
Merge changes from github.
|
2016-09-14 00:17:57 -07:00 |
|
data_flow_grad.py
|
Rename NoGradient -> NotDifferentiable, to make it clear
|
2016-09-09 12:17:24 -07:00 |
|
data_flow_ops.py
|
Fix spelling in docstrings.
|
2016-09-28 22:18:58 -07:00 |
|
embedding_ops.py
|
Fix broken lists in documentation, and back-quote more identifiers.
|
2016-09-28 23:50:16 -07:00 |
|
functional_ops.py
|
Switch ops in functional_ops, math_ops, state_ops, string_ops,
|
2016-09-09 10:31:16 -07:00 |
|
gradient_checker_test.py
|
Use np.maximum in gradient_checker. Previously, max was used, but max(x, nan)
|
2016-08-30 13:02:57 -07:00 |
|
gradient_checker.py
|
Use np.maximum in gradient_checker. Previously, max was used, but max(x, nan)
|
2016-08-30 13:02:57 -07:00 |
|
gradients_test.py
|
Cleanup
|
2016-09-22 09:47:55 -07:00 |
|
gradients.py
|
Removes a few limitation of Defun:
|
2016-09-08 17:17:45 -07:00 |
|
hidden_ops.txt
|
Add SpaceToBatchND and BatchToSpaceND ops for > 2 spatial dimensions.
|
2016-09-21 09:17:53 -07:00 |
|
histogram_ops_test.py
|
Update copyright for 3p/tf/python.
|
2016-06-02 13:38:16 -07:00 |
|
histogram_ops.py
|
Deprecate op_scope by extending name_scope and move all callers to the new name_scope.
|
2016-08-09 12:03:16 -07:00 |
|
image_grad_test.py
|
To make the tests run both on GPU and CPU, when available, override use_gpu to
|
2016-09-09 23:48:49 -07:00 |
|
image_grad.py
|
Switch to C++ shape inference fns for more ops in image_grad, io_ops,
|
2016-09-07 16:47:39 -07:00 |
|
image_ops_test.py
|
Internal change.
|
2016-09-15 10:33:16 -07:00 |
|
image_ops.py
|
Fix broken lists in documentation, and back-quote more identifiers.
|
2016-09-28 23:50:16 -07:00 |
|
init_ops.py
|
Avoid errors when initializing zero-size variables with get_variable
|
2016-09-22 09:18:36 -07:00 |
|
io_ops.py
|
TF Checkpoint V2: SaveV2, RestoreV2, MergeV2Checkpoints ops.
|
2016-09-23 16:18:50 -07:00 |
|
linalg_grad.py
|
Add gradient for self_adjoint_eig op. Make the unit test sharded.
|
2016-09-13 16:03:32 -07:00 |
|
linalg_ops.py
|
Add gradient for self_adjoint_eig op. Make the unit test sharded.
|
2016-09-13 16:03:32 -07:00 |
|
logging_ops.py
|
Rename NoGradient -> NotDifferentiable, to make it clear
|
2016-09-09 12:17:24 -07:00 |
|
math_grad_test.py
|
To make the tests run both on GPU and CPU, when available, override use_gpu to
|
2016-09-09 23:48:49 -07:00 |
|
math_grad.py
|
Consolidate fft ops: Replace fft* ops with their batch_fft* equivalents and deprecate the latter.
|
2016-09-14 17:02:25 -07:00 |
|
math_ops_test.py
|
Clean up another batch of tensorflow tests that are using use_gpu.
|
2016-09-09 23:47:03 -07:00 |
|
math_ops.py
|
Update trace, matrix_set_diag, matrix_diag_part and their gradients to work for rectangular matrices.
|
2016-09-29 13:35:30 -07:00 |
|
nn_batchnorm_test.py
|
Add weighted_moments, and allow batch norm to use it to compute frequency-weighted statistics.
|
2016-09-29 15:39:40 -07:00 |
|
nn_fused_batchnorm_test.py
|
Add a fused batch normalization op, supporting both cuDNN and an Eigen-based
|
2016-09-27 18:08:38 -07:00 |
|
nn_grad.py
|
Add a fused batch normalization op, supporting both cuDNN and an Eigen-based
|
2016-09-27 18:08:38 -07:00 |
|
nn_ops.py
|
Add syntax highlighting and fix some lists in docs.
|
2016-09-28 23:35:56 -07:00 |
|
nn_test.py
|
Implement the Concatenated ReLU (CReLu) non-linearity. This non-linearity makes it possible to reduce the depth of lower layers in a convolution tower by a factor of two at no cost in accuracy. See: http://arxiv.org/abs/1603.05201
|
2016-08-23 17:05:19 -07:00 |
|
nn_xent_test.py
|
Use 32bit floats to compute cross entropies. 16 bit floats aren't accurate
|
2016-07-22 14:04:04 -07:00 |
|
nn.py
|
Add weighted_moments, and allow batch norm to use it to compute frequency-weighted statistics.
|
2016-09-29 15:39:40 -07:00 |
|
numerics.py
|
Deprecate op_scope by extending name_scope and move all callers to the new name_scope.
|
2016-08-09 12:03:16 -07:00 |
|
parsing_ops.py
|
Rename NoGradient -> NotDifferentiable, to make it clear
|
2016-09-09 12:17:24 -07:00 |
|
partitioned_variables.py
|
Remove unnecessary imports in some python files.
|
2016-09-19 10:47:47 -07:00 |
|
random_ops.py
|
Rename NoGradient -> NotDifferentiable, to make it clear
|
2016-09-09 12:17:24 -07:00 |
|
rnn_cell.py
|
Fix broken lists in documentation, and back-quote more identifiers.
|
2016-09-28 23:50:16 -07:00 |
|
rnn.py
|
Documentation fix: markdown requires an empty line before ``` block.
|
2016-09-29 12:26:21 -07:00 |
|
script_ops.py
|
Re-establish ability for py_func to accept Tout as tuple as well as list.
|
2016-09-22 08:35:15 -07:00 |
|
sdca_ops.py
|
Move SDCA optimizer kernels to core.
|
2016-09-27 14:18:10 -07:00 |
|
seq2seq.py
|
Merge changes from github.
|
2016-09-09 17:19:12 -07:00 |
|
session_ops.py
|
Fix a doube free bug in persistent tensor. Also improved delete_session_tensor, and added a delete method to TensorHandle class.
|
2016-06-09 11:48:47 -07:00 |
|
sparse_grad.py
|
Rename NoGradient -> NotDifferentiable, to make it clear
|
2016-09-09 12:17:24 -07:00 |
|
sparse_ops.py
|
Quote tf.* identifiers in documentation.
|
2016-09-29 00:50:02 -07:00 |
|
special_math_ops_test.py
|
Merge changes from github.
|
2016-09-21 14:31:57 -07:00 |
|
special_math_ops.py
|
Merge changes from github.
|
2016-09-21 14:31:57 -07:00 |
|
standard_ops.py
|
Add scatter_mul and scatter_div state ops for CPU and GPU.
|
2016-09-08 04:18:22 -07:00 |
|
state_grad.py
|
Rename NoGradient -> NotDifferentiable, to make it clear
|
2016-09-09 12:17:24 -07:00 |
|
state_ops.py
|
Move initialize_all_tables to Variables documentation.
|
2016-09-28 20:56:04 -07:00 |
|
string_ops.py
|
Remove unnecessary imports in some python files.
|
2016-09-19 10:47:47 -07:00 |
|
summary_ops.py
|
Rename NoGradient -> NotDifferentiable, to make it clear
|
2016-09-09 12:17:24 -07:00 |
|
template.py
|
Fix symbol names in documentation.
|
2016-09-28 21:35:59 -07:00 |
|
tensor_array_grad.py
|
Rename NoGradient -> NotDifferentiable, to make it clear
|
2016-09-09 12:17:24 -07:00 |
|
tensor_array_ops.py
|
Add TensorArray gather and scatter. Deprecate old pack and unpack ops.
|
2016-09-08 17:33:12 -07:00 |
|
variable_scope.py
|
Quote tf.* identifiers in documentation.
|
2016-09-29 00:50:02 -07:00 |
|
variables.py
|
Update documentation script to not omit operators
|
2016-09-12 17:17:50 -07:00 |
